CN115292933B

CN115292933B - Geographical weighted regression model creation method and device for analyzing correlation of ore-forming elements

Info

Publication number: CN115292933B
Application number: CN202210937205.7A
Authority: CN
Inventors: 王文磊; 袁长江; 乔东旭; 王文君
Original assignee: INSTITUTE OF GEOMECHANICS CHINESE ACADEMY OF GEOLOGICAL SCIENCES
Current assignee: INSTITUTE OF GEOMECHANICS CHINESE ACADEMY OF GEOLOGICAL SCIENCES
Priority date: 2022-08-05
Filing date: 2022-08-05
Publication date: 2023-02-17
Anticipated expiration: 2042-08-05
Also published as: CN115292933A

Abstract

The application provides a method and a device for establishing a geographical weighted regression model for analyzing correlation of ore-forming elements, wherein the establishing method comprises the following steps: taking any one of a plurality of sampling points of a target research area as a regression analysis point, and constructing a plurality of ellipse search areas taking the regression analysis point as an ellipse center; determining a target ellipse searching area of the regression analysis point according to a space weighting U statistic method; acquiring all sampling points included in a target ellipse searching area; determining a plurality of regression coefficients according to the predetermined target weight from each sampling point to the regression analysis point; and creating a corresponding geographical weighted regression model according to the regression coefficients. By adopting the technical scheme provided by the application, the information extraction effect of the corresponding preset mineralization elements in the mining area can be improved, the method can also be used for quantitatively analyzing and describing the symbiotic relationship between the preset mineralization elements and the preset geological activity sensitive elements in the mining area, and the applicability of the geographic weighted regression model is also improved.

Description

Geographical weighted regression model creation method and device for analyzing correlation of ore-forming elements

Technical Field

The application relates to the field of geospatial data distribution pattern analysis, in particular to a method and a device for establishing a geographic weighted regression model for analyzing correlation of mine-forming elements.

Background

The geochemical data is multivariable data, each sampling point has concentration value information of a plurality of elements, and due to the complexity of geological processes, the multi-stage superposition of an ore forming process and the mutual influence and restriction among geochemical elements, the geochemical abnormal information is a result of multi-element comprehensive indexes and multi-factor composite superposition. For example, the elemental concentration distributions of survey geochemical data are characterized by anisotropy and non-uniformity as a result of the influence of controlling factors such as geological formations, magma, etc. By analyzing the anisotropy of each element, the spatial distribution of each element is explored, the correlation of each element under different geological conditions is further excavated, and the influence factors of each ore formation are analyzed, so that the method is favorable for determining the abnormal information of the ore formation so as to better find the ore.

At present, a method for researching and considering the anisotropy of geological information is to improve a circular bandwidth in a traditional geographical weighted regression model into an elliptical bandwidth, and the method has more advantages in analyzing the anisotropy characteristics of various ore forming influence factors in the field of ore forming prediction compared with the traditional geographical weighted regression model, but a bandwidth optimal selection method commonly used in the process of screening sample data in an elliptical search area, such as a cross validation method, an akage pool information quantity criterion and the like, cannot realize the fine classification of geochemical data abnormality and background. And secondly, the optimal ellipse of the geochemical sampling point is obtained by using a spatial weighting U statistic method, the spatial weighting U statistic method has good classification characteristics, is mainly used for anisotropic analysis of geochemical data, and can effectively reduce the separation error of anomaly and background, but the spatial weighting U statistic method is limited by the classification characteristics of a calculation model, and is difficult to be directly applied to various data types. Therefore, how to better analyze the relationship between ore-forming elements so as to determine abnormal information causing ores becomes a problem to be urgently solved.

Disclosure of Invention

In view of this, an object of the present application is to provide a method and an apparatus for creating a geographic weighted regression model for analyzing correlation of mineral-forming elements, which can determine a target ellipse search region by using a spatial weighted U statistic method, construct a geographic weighted regression model according to the target ellipse search region and a predetermined target weight, provide a new method for analyzing anisotropy of geospatial data, not only can improve information extraction effect of corresponding predetermined mineral-forming elements in a mining area, and achieve maximum separation of background and anomaly in the ellipse search region, but also can be used for quantitatively analyzing and describing a symbiotic relationship between the predetermined mineral-forming elements and predetermined activity sensitive elements in the mining area, expand application scenarios of spatial U statistic methods, and improve applicability of the geographic weighted regression model.

The application mainly comprises the following aspects:

in a first aspect, an embodiment of the present application provides a method for creating a geo-weighted regression model for analyzing correlation of mine-forming elements, where the method for creating the geo-weighted regression model includes:

acquiring sampling point information of a region corresponding to a plurality of sampling points in a target research region; the sampling point information comprises spatial position coordinates of a region corresponding to the sampling point, the concentration of a preset mineralization element and the concentration of at least one preset geological activity sensitive element;

taking any one of the sampling points as a regression analysis point, and constructing a plurality of ellipse search areas taking the regression analysis point as an ellipse center according to predetermined ellipse search parameters;

determining a target ellipse searching region of the regression analysis point in a plurality of ellipse searching regions of the regression analysis point according to a space weighting U statistical method;

acquiring all sampling points included in the target ellipse searching area;

determining a plurality of regression coefficients corresponding to the regression analysis point in the target ellipse search area according to sampling point information of a region corresponding to each sampling point in the target ellipse search area and a predetermined target weight from each sampling point to the regression analysis point;

and creating a geographical weighted regression model corresponding to the target ellipse searching region with the regression analysis point as the ellipse center according to the regression coefficient, so as to be used for analyzing the influence degree of the concentration of each preset geological activity sensitive element in the region corresponding to each sampling point in the target ellipse searching region with the regression analysis point as the ellipse center on the corresponding target mineralization element.

Further, the ellipse search parameter is determined by:

determining a long half shaft range and a short half shaft range of an ellipse search area according to the spatial position coordinates of a region corresponding to each sampling point;

determining a value range of an equivalent radius of the ellipse search area according to the range of the major axis and the range of the minor axis of the ellipse search area;

splitting the value range of the equivalent radius according to the step length and determining a plurality of equivalent radii according to the value range of the equivalent radius and the preset step length;

splitting the value range of the compression coefficient according to the preset compression coefficient interval according to the value range of the preset compression coefficient and the preset compression coefficient interval, and determining a plurality of compression coefficients;

splitting the value range of the azimuth angle according to a preset azimuth interval and a preset azimuth interval to determine a plurality of azimuth angles;

determining the plurality of equivalent radii, the plurality of compression coefficients, and the plurality of azimuth angles as ellipse search parameters.

Further, the step of constructing a plurality of ellipse search regions with the regression analysis point as the center of the ellipse according to the predetermined ellipse search parameters includes:

according to a predetermined ellipse search parameter, acquiring a plurality of equivalent radii, a plurality of compression coefficients and a plurality of azimuth angles in the ellipse parameter;

in the equivalent radiuses, taking the initial equivalent radius as the current equivalent radius, and constructing an ellipse search area corresponding to each azimuth angle of the compression coefficient under the current equivalent radius by taking the regression analysis point as the ellipse center aiming at each compression coefficient;

determining whether there is a next equivalent radius;

if so, updating the next equivalent radius to the current equivalent radius, and continuously constructing a plurality of ellipse search areas with the next equivalent radius as the current equivalent radius;

and if not, obtaining a plurality of ellipse searching areas corresponding to each equivalent radius in all equivalent radii taking the regression analysis point as the center of the ellipse.

Further, the step of determining a target elliptical search area of the regression analysis point from among the plurality of elliptical search areas of the regression analysis point includes:

in a plurality of ellipse searching areas of the regression analysis point, aiming at each equivalent radius in the ellipse searching parameters, determining a spatial weighting U statistic corresponding to each ellipse searching area in the plurality of ellipse searching areas corresponding to the equivalent radius;

in the space weighting U statistic corresponding to each ellipse searching area under the equivalent radius, determining the ellipse searching area with the largest absolute value of the space weighting U statistic as the optimal ellipse searching area of the equivalent radius;

and sequencing the absolute values of the spatial weighting U statistics corresponding to the optimal ellipse searching region of each equivalent radius according to each equivalent radius in the ellipse searching parameters, and determining the optimal ellipse searching region corresponding to the spatial weighting U statistics with the maximum absolute value as the target ellipse searching region of the regression analysis point.

Further, the spatial weighting U statistic corresponding to each elliptical search area is determined by the following steps:

aiming at each oval search area, acquiring the spatial position coordinates of the area corresponding to each sampling point in all the sampling points in the oval search area and the concentration of the preset mineralization elements;

determining a major semi-axis and a minor semi-axis of the ellipse search area according to the ellipse parameters corresponding to the ellipse search area;

determining the target weight from each sampling point to the regression analysis point according to the major axis and the minor axis of the ellipse search area and the spatial position coordinates of the area corresponding to each sampling point in all the sampling points in the ellipse search area;

and determining the spatial weighting U statistic corresponding to the elliptic search area according to the concentration of the predetermined ore-forming elements in the area corresponding to each sampling point and the target weight corresponding to each sampling point.

Further, the step of determining the target weight from each sampling point to the regression analysis point according to the semi-major axis and the semi-minor axis of the ellipse search area and the spatial position coordinates of the area corresponding to each sampling point in all the sampling points in the ellipse search area includes:

determining a weight parameter corresponding to each sampling point according to the major axis and the minor axis of the ellipse searching region and the spatial position coordinates of a region corresponding to each sampling point in all the sampling points in the ellipse searching region;

for each sampling point, determining whether the weight parameter corresponding to the sampling point is in a range greater than zero and less than or equal to one;

if so, squaring the weight parameter corresponding to the sampling point to obtain a first weight parameter, and squaring the difference between the first weight parameter and the first weight parameter to obtain the target weight from the sampling point to the regression analysis point;

if not, determining zero as the target weight from the sampling point to the regression analysis point.

Further, the step of determining a plurality of regression coefficients corresponding to the regression analysis point in the target ellipse search region according to the sampling point information of the area corresponding to each sampling point in the target ellipse search region and the predetermined target weight from each sampling point to the regression analysis point includes:

determining a column vector consisting of the concentrations of the predetermined mineralizing elements in the area corresponding to each sampling point in the target ellipse search area as a dependent variable column vector of a geographical weighted regression model according to the concentrations of the predetermined mineralizing elements in the sampling point information of the area corresponding to each sampling point in the target ellipse search area;

determining a matrix formed by the concentrations of at least one preset geological activity sensitive element in the area corresponding to each sampling point in the target ellipse search area as an independent variable matrix of a geographical weighted regression model according to the concentration of at least one preset geological activity sensitive element in the sample point information of the area corresponding to each sampling point;

determining a diagonal matrix formed by the target weights from each sampling point to the regression analysis point as a spatial weight matrix of a geographical weighting regression model according to the predetermined target weights from each sampling point to the regression analysis point in the target ellipse search area;

multiplying the transpose matrix of the independent variable matrix, the space weight matrix and the independent variable matrix to obtain a first matrix;

multiplying the inverse matrix of the first matrix by the transposed matrix of the independent variable matrix to obtain a second matrix;

multiplying the second matrix by the space weight matrix to obtain a third matrix;

multiplying the third matrix by the dependent variable column vector to obtain a regression coefficient column vector;

and determining a plurality of regression coefficients corresponding to the regression analysis points in the target ellipse search area according to the regression coefficient column vectors.

In a second aspect, an embodiment of the present application further provides a device for creating a geo-weighted regression model for analyzing correlation of mine-forming elements, where the device for creating the geo-weighted regression model includes:

the first acquisition module is used for acquiring sampling point information of a region corresponding to a plurality of sampling points in a target research region; the sampling point information comprises spatial position coordinates of a region corresponding to the sampling point, the concentration of a preset mineralizing element and the concentration of at least one preset geological activity sensitive element;

the construction module is used for constructing a plurality of ellipse search areas which take any one of the plurality of sampling points as a regression analysis point and take the regression analysis point as an ellipse center according to predetermined ellipse search parameters;

the first determining module is used for determining a target ellipse searching region of the regression analysis point in a plurality of ellipse searching regions of the regression analysis point according to a space weighting U statistical method;

the second acquisition module is used for acquiring all sampling points in the target ellipse search area;

the processing module is used for determining a plurality of regression coefficients corresponding to the regression analysis point in the target ellipse search area according to sampling point information of a region corresponding to each sampling point in the target ellipse search area and a predetermined target weight from each sampling point to the regression analysis point;

and the creating module is used for creating a geographical weighting regression model corresponding to the target ellipse searching region with the regression analysis point as the center of the ellipse according to the regression coefficient, and is used for analyzing the influence degree of the concentration of each preset geological activity sensitive element in the region corresponding to each sampling point in the target ellipse searching region with the regression analysis point as the center of the ellipse on the corresponding target mineral forming element.

In a third aspect, an embodiment of the present application further provides an electronic device, including: a processor, a memory and a bus, the memory storing machine-readable instructions executable by the processor, the processor and the memory communicating over the bus when the electronic device is operating, the machine-readable instructions, when executed by the processor, performing the steps of the method for creating a geo-weighted regression model for analyzing correlations of mineralised elements as described above.

In a fourth aspect, embodiments of the present application further provide a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, performs the steps of the above-mentioned geographic weighted regression model creation method for analyzing correlations of mineralogical elements.

The embodiment of the application provides a method and a device for establishing a geographical weighted regression model for analyzing correlation of ore-forming elements, wherein the establishing method comprises the following steps: acquiring sampling point information of a region corresponding to a plurality of sampling points in a target research region; the sampling point information comprises spatial position coordinates of a region corresponding to the sampling point, the concentration of a preset mineralizing element and the concentration of at least one preset geological activity sensitive element; taking any one of the sampling points as a regression analysis point, and constructing a plurality of ellipse search areas taking the regression analysis point as an ellipse center according to predetermined ellipse search parameters; determining a target ellipse searching region of the regression analysis point in a plurality of ellipse searching regions of the regression analysis point according to a space weighting U statistical method; acquiring all sampling points included in the target ellipse searching area; determining a plurality of regression coefficients corresponding to the regression analysis point of the target ellipse search area according to the sampling point information of the area corresponding to each sampling point in the target ellipse search area and the predetermined target weight from each sampling point to the regression analysis point; and creating a geographical weighted regression model corresponding to the target ellipse searching region with the regression analysis point as the ellipse center according to the regression coefficient, so as to be used for analyzing the influence degree of the concentration of each preset geological activity sensitive element in the region corresponding to each sampling point in the target ellipse searching region with the regression analysis point as the ellipse center on the corresponding target mineralization element.

Therefore, the technical scheme provided by the application can select a spatial weighting U statistic method to determine a target elliptical search area, a geographical weighting regression model is constructed according to the target elliptical search area and the predetermined target weight, a new method is provided for analyzing the anisotropy of the geospatial data, the method can be effectively applied to geological application scenes of the non-stationarity and anisotropy analysis of geochemical data space, the extraction effect of classification information of the background and abnormal distribution of the corresponding predetermined mineral forming elements in a mining area can be improved, the maximum separation of the background and the abnormality in the elliptical search area is realized, the method can also be used for quantitatively analyzing and describing the symbiotic relationship between the predetermined mineral forming elements and the predetermined sensitive elements to geological activity in the mining area, the application scene of the spatial U statistic method is expanded, and the applicability of the geographical weighting regression model is also improved.

In order to make the aforementioned objects, features and advantages of the present application more comprehensible, preferred embodiments accompanied with figures are described in detail below.

Drawings

To more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and those skilled in the art can also obtain other related drawings based on the drawings without inventive efforts.

FIG. 1 is a flow chart illustrating a method for creating a geo-weighted regression model for analyzing relevance of mineralogical elements according to an embodiment of the present application;

FIG. 2 is a flow chart illustrating another method for creating a geo-weighted regression model for analyzing correlation of mineralizing elements, provided by an embodiment of the present application;

FIG. 3 illustrates a flow diagram of conventional geo-weighted regression model parameter estimation;

FIG. 4 is a diagram illustrating a projection of a weight kernel function of a conventional geo-weighted regression model;

FIG. 5 is a diagram illustrating kernel function fixed and adaptive bandwidth types for a conventional geo-weighted regression model;

FIG. 6 illustrates a workflow diagram of an existing anisotropy-based refined geoweighted regression model;

FIG. 7 shows a schematic diagram of a prior art circular, oval-shaped local search window;

FIG. 8 is a graph showing the variation of the prior U value with the equivalent radius;

FIG. 9 illustrates an equivalent radius R provided by an embodiment of the present application ₁ -compression factor α ₁ A schematic view of each azimuth of a plurality of elliptical search areas;

FIG. 10 is a flow chart illustrating the operation of a spatially weighted U statistic based improved geo-weighted regression model provided by an embodiment of the present application;

FIG. 11 illustrates a work flow of a spatial non-stationarity analysis of geochemical data of a mine according to an embodiment of the present application;

FIG. 12 illustrates a graph of interpolation of predetermined mineralizing element concentration observations as provided by an embodiment of the present application;

figure 13 shows an interpolation plot of predicted mineralizing element concentration estimates based on a conventional GWR model as provided by an embodiment of the present application;

figure 14 shows an interpolation plot of the estimated value of the concentration of predetermined mineralizing elements for the improved GWR model based on spatial U statistics as provided by an embodiment of the present application;

figure 15 illustrates conventional GWR model-based regression coefficients β provided by embodiments of the present application ₀ A result interpolation graph;

figure 16 illustrates the improved GWR model regression coefficient β based on spatial U statistics as provided by embodiments of the present application ₀ A result interpolation graph;

figure 17 illustrates conventional GWR model-based regression coefficients β provided by embodiments of the present application ₁ A result interpolation graph;

figure 18 illustrates the improved GWR model regression coefficient β based on spatial U statistics as provided by embodiments of the present application ₁ A result interpolation graph;

fig. 19 illustrates a local R based on a conventional GWR model provided in an embodiment of the present application ² A result interpolation graph;

figure 20 illustrates improving GWR model local R based on spatial U statistics as provided by embodiments of the present application ² A result interpolation graph;

figure 21 illustrates a space U statistic based improved GWR model multi-factor regression local R provided by embodiments of the application ² A result interpolation graph;

figure 22 shows a graph of multi-factor regression residual result interpolation based on spatial U statistic improved GWR model provided by embodiments of the present application;

fig. 23 is a block diagram of a geographic weighted regression model creation apparatus for analyzing correlation of mine-forming elements according to an embodiment of the present disclosure;

fig. 24 is a second block diagram of a geographic weighted regression model creation apparatus for analyzing correlation of mine-forming elements according to an embodiment of the present application;

fig. 25 shows a schematic structural diagram of an electronic device according to an embodiment of the present application.

Detailed Description

In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it should be understood that the drawings in the present application are for illustrative and descriptive purposes only and are not intended to limit the scope of the present application. Further, it should be understood that the schematic drawings are not drawn to scale. The flowcharts used in this application illustrate operations implemented according to some embodiments of the present application. It should be understood that the operations of the flow diagrams may be performed out of order, and that steps without logical context may be performed in reverse order or concurrently. In addition, one skilled in the art, under the guidance of the present disclosure, may add one or more other operations to the flowchart, or may remove one or more operations from the flowchart.

In addition, the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. The components of the embodiments of the present application, generally described and illustrated in the figures herein, can be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present application, presented in the accompanying drawings, is not intended to limit the scope of the claimed application, but is merely representative of selected embodiments of the application. All other embodiments, which can be obtained by a person skilled in the art without making any inventive step based on the embodiments of the present application, fall within the scope of protection of the present application.

To enable those skilled in the art to use the present disclosure in connection with a particular application scenario "creation of a geoweighted regression model for analysis of mineralogical element correlations," the following embodiments are given, and it will be apparent to those skilled in the art that the general principles defined herein may be applied to other embodiments and application scenarios without departing from the spirit and scope of the present application.

The method, apparatus, electronic device or computer-readable storage medium described in the embodiments of the present application may be applied to any scenario that requires creating a geographic weighted regression model for analyzing correlation of ore-forming elements, and the embodiments of the present application are not limited to a specific application scenario.

The background and prior art of the solution is described herein for a better understanding of the present application.

First, changes in geographic location in spatial analysis can cause changes in the relationships or structure between model variables, i.e., spatial non-stationarity. In a traditional geographical Weighted Regression model, a specific parameter estimation process refers to fig. 3, fig. 3 is a flow chart of parameter estimation of the traditional geographical Weighted Regression model, and as shown in fig. 3, the step of parameter estimation includes 1) inputting all data; 2) Selecting an analysis point x in all data centers ₁ (ii) a 3) Setting initial value bandwidth b = r of search ₁ (ii) a 4) Selecting sub-sample data through the bandwidth b; 5) Computing weights w for subsample data from kernel functions _ij (ii) a 6) Estimating parameters of the analysis points by using a weighted least square regression method; 7) Calculating an Akaba information criterion AIC value of the current analysis point; 8) Varying bandwidth b, b = r ₁ ，r ₂ ，…，r _n Judging whether the bandwidth b reaches the constraint condition, if not, entering 2) reselecting an analysis point, if so, entering 9) comparing AIC values, and selecting the bandwidth corresponding to the minimum AIC value as the optimal bandwidth b ^* ＝r _j (ii) a 10 Determining whether the bandwidth is a fixed bandwidth, if so, entering 11) according to the optimal bandwidth b ^* Traversing the whole area and repeating the steps 4) to 6); if not, entering 12) according to the optimal bandwidth b ^* Repeating the above steps 4) to 6), calculating the analysis point x ₁ Regression parameters of anda regression equation; 13 ) traverse the whole area, repeating the above steps 3) to 12).

Secondly, the "weight" in the traditional geographic weighted regression model refers to a distance function from the geographic spatial position of the regression point to the geographic spatial positions of other observation points, each spatial weight value is obtained through calculation of a weight kernel function, the most common weight kernel functions in practical application are a gaussian function and a quadratic function, please refer to fig. 4, fig. 4 is a projection diagram of the weight kernel function of the traditional geographic weighted regression model, and as shown in fig. 4, the projections of the gaussian function and the quadratic function on a two-dimensional plane are both circular; as shown in fig. 5, fig. 5 is a schematic diagram of kernel function fixed bandwidth and adaptive bandwidth types of a traditional geographic weighted regression model, where a fixed distance (distance threshold b) is used when local geographic weighted analysis is performed on the fixed bandwidth, and an adaptive bandwidth determines the size of b by using the number k of adjacent points (sample points) where regression analysis points are fixed; selecting a proper bandwidth as a necessary node for solving a GWR model, wherein a common method is a cross validation method (CV) and an Akabane Information Criterion (AIC), and selecting an optimal bandwidth value by minimizing the CV or the AIC value; under the conditions of different bandwidth types, obtaining results such as regression parameters, regression equations and the like by using a least square regression method according to the optimal bandwidth so as to quantitatively describe the spatial non-stationary relation between the dependent variable and the independent variable; and in the aspect of model evaluation, the multi-purpose fitting goodness, the fitting goodness adjustment and the AIC value are used for evaluating the parameter estimation effect, and the reliability of the model result is measured.

Finally, geospatial data not only has spatially non-stationary characteristics, but also often exhibits differences in physicochemical properties and the like, i.e., anisotropy, in different directions. For example, the elemental concentration distributions of survey geochemical data are characterized by anisotropy and non-uniformity as a function of controlling factors such as geological formations, magma, and the like. At present, the researchers have studied and designed a geography-weighted regression method considering the anisotropy of the geography information, please refer to fig. 6, fig. 6 is a work flow chart of the improved geography-weighted regression model based on the anisotropy, as shown in fig. 6, the steps include: 1) Selecting an estimated sampling point i, and defining a residual C =0; 2) Defining an initial elliptical window (a) ₁ ，b ₁ ，θ ₁ ) (ii) a 3) Screening sample points according to the oval window; 4) Computing weights w using a joint probability density function _i (ii) a 5) Calculating weighted least square method estimation parameter and residual value c _i And determining whether the residual is equal to zero and greater than the residual value, i.e., whether C = 0/C > C is satisfied _i (ii) a If not, then 6) assigning the residual value to the residual, i.e. C = C _i (ii) a If yes, 7) changing the ellipse shape, and establishing a series of ellipse windows: (a) ₂ ，b ₁ ，θ ₁ )，(a ₁ ，b ₂ ，θ ₁ )，(a ₁ ，b ₁ ，θ ₂ )…，(a _i ，b _j ，θ _k ) And determining a _i ，b _j ，θ _k Whether a constraint condition is reached; if yes, 8) obtaining an optimal ellipse of the sampling point i and a corresponding your sum equation; if not, entering 3) re-screening the sample points; 9) Repeating the steps 1) to 8), moving to the whole area, and obtaining the optimal ellipse and the your sum equation of each estimated sample point; the method is to improve the circular bandwidth in the traditional geographic weighted regression model into the elliptical bandwidth. Referring to fig. 7, fig. 7 is a schematic diagram of a conventional circular and elliptical local search window, and as shown in fig. 7, three bandwidth parameters are respectively a short axis (a), a long axis (b) and a rotation angle (θ) of an ellipse, an elliptical weight kernel function model is developed, and an optimal ellipse is selected by using a method of minimum residual square sum, which is more advantageous than a conventional model in analyzing anisotropic characteristics of each mineralization influencing factor in the field of mineralization prediction. In addition, in the prior art, an optimal ellipse of a geochemical sampling point is obtained by using a spatial U statistical method, please refer to fig. 8, fig. 8 is a graph of a change curve of a current U value along with an equivalent radius, as shown in fig. 8, a vertical axis U is a U statistical quantity, a horizontal axis R is an equivalent radius, and it is considered that | U $ zero _max All the sampling points in the determined ellipse have the strongest spatial correlation, the sampling points in the optimal ellipse can belong to the same distribution (abnormal or background), and all parameters of the local optimal ellipse, namely the equivalent radius (R), the compression coefficient (alpha) and the major axis azimuth angle (theta), can be obtainedLocal anisotropy characteristics of the geochemical elements in the distribution are represented, and background and anomaly in the geochemical data are effectively separated, so that the mineralization anomaly information is extracted.

It is worth noting that the geographic weighted regression model has been widely applied in the fields of social economy, ecological environment, remote sensing, earth science and the like. Because the weight kernel function used in the traditional geographic weighting model has isotropy, the weight kernel function is circular projection in a two-dimensional space, the weight values of data at the same distance from a central analysis point are the same, when the geochemical data is analyzed and surveyed, the characteristic that the distribution rule of geochemical elements has anisotropy is ignored, the spatial distribution of complex geological processes and products thereof cannot be objectively reflected, and the bandwidth optimal selection method commonly used when the sample data of an ellipse search area is screened, such as a cross validation method, an akage information content criterion and the like, cannot realize the fine classification of geochemical data abnormity and background. The spatial weighting U statistic method has good classification characteristics, is mainly used for anisotropic local singularity analysis of geochemical data, can effectively reduce separation errors of abnormalities and backgrounds, and can also be used for remote sensing abnormal information extraction of a two-dimensional spectral feature space. Therefore, how to better analyze the relationship between the ore-forming elements so as to determine the abnormal information causing the ore becomes a problem to be solved urgently.

Based on this, the application provides a method and a device for creating a geographical weighted regression model for analyzing correlation of ore-forming elements, wherein the creating method comprises the following steps: acquiring sampling point information of a region corresponding to a plurality of sampling points in a target research region; the sampling point information comprises spatial position coordinates of a region corresponding to the sampling point, the concentration of a preset mineralization element and the concentration of at least one preset geological activity sensitive element; taking any one of the sampling points as a regression analysis point, and constructing a plurality of ellipse search areas taking the regression analysis point as an ellipse center according to predetermined ellipse search parameters; determining a target ellipse searching region of the regression analysis point in a plurality of ellipse searching regions of the regression analysis point according to a space weighting U statistical method; acquiring all sampling points included in the target ellipse searching area; determining a plurality of regression coefficients corresponding to the regression analysis point of the target ellipse search area according to the sampling point information of the area corresponding to each sampling point in the target ellipse search area and the predetermined target weight from each sampling point to the regression analysis point; and creating a geographical weighted regression model corresponding to the target ellipse searching region with the regression analysis point as the ellipse center according to the regression coefficient, so as to be used for analyzing the influence degree of the concentration of each preset geological activity sensitive element in the region corresponding to each sampling point in the target ellipse searching region with the regression analysis point as the ellipse center on the corresponding target mineralization element.

For the purpose of facilitating an understanding of the present application, the technical solutions provided in the present application will be described in detail below with reference to specific embodiments.

Referring to fig. 1, fig. 1 is a flowchart of a method for creating a geo-weighted regression model for analyzing correlation of mine-forming elements according to an embodiment of the present application, as shown in fig. 1, the method includes:

s101, acquiring sampling point information of a region corresponding to a plurality of sampling points in a target research area;

in the step, the sampling point information comprises the spatial position coordinates of the area corresponding to the sampling point, the concentration of the preset mineralizing elements and the concentration of at least one preset geological activity sensitive element.

Illustratively, the target research area is a mining area, the sampling points can be 1. Here, the background refers to a region containing no ore or not affected by mineralization, and the values of elements in the region are normal and vary within a certain range, for example, the concentration of the predetermined ore-forming elements in the region does not exceed the preset background and abnormal boundary threshold, which indicates that the predetermined ore-forming elements are in the background region, and the plastid in the background region does not have special geochemical characteristics. Geochemical anomalous zones, in turn, often exhibit significantly different geochemical characteristics and have anomalous values, such as concentrations of predetermined mineralizing elements within the zone exceeding predetermined background and anomalous cutoff thresholds. The identification and isolation of anomalies and backgrounds is a fundamental problem in the field of exploration geochemistry. The mineralization is the superposition of various information, and the spatial distribution of the geochemical data elements is the result of coupling of various factors in time and space, so that the mineralization abnormal information is a good geochemical mineral exploration mark.

S102, taking any one of the plurality of sampling points as a regression analysis point, and constructing a plurality of ellipse search areas taking the regression analysis point as an ellipse center according to a predetermined ellipse search parameter;

it should be noted that, please refer to fig. 2, fig. 2 is a flowchart of another method for creating a geo-weighted regression model for analyzing correlation of mineralization elements according to an embodiment of the present application, and as shown in fig. 2, the elliptical search parameters are determined through the following steps:

s201, determining a long half shaft range and a short half shaft range of an ellipse search area according to the spatial position coordinates of the area corresponding to each sampling point;

s202, determining a value range of an equivalent radius of the ellipse search area according to a major semiaxis range and a minor semiaxis range of the ellipse search area;

s203, splitting the value range of the equivalent radius according to the step length according to the value range of the equivalent radius and the preset step length to determine a plurality of equivalent radii;

s204, splitting the value range of the compression coefficient according to the preset compression coefficient interval according to the preset value range of the compression coefficient and the preset compression coefficient interval, and determining a plurality of compression coefficients;

s205, splitting the value range of the azimuth angle according to a preset azimuth interval and a preset azimuth interval to determine a plurality of azimuth angles;

s206, determining the equivalent radiuses, the compression coefficients and the azimuth angles as ellipse searching parameters.

In the step, when the optimal ellipse window is selected by the space weighting U statistic method, an ellipse search area is constructed by ellipse search parameters, and an ellipse search area B is constructed by the ellipse search parameters _x (R, alpha, theta) is based on three parameters, and the equivalent radius R, the compression coefficient alpha and the azimuth angle theta are ellipse searching parameters, wherein the equivalent radius

The compression coefficient α = b/a, θ is the azimuth; here, a is a major semi-axis of the ellipse search region, b is a minor semi-axis of the ellipse search region, and the azimuth angle θ is an angle between the major semi-axis of the ellipse search region and a coordinate axis in the east-right direction in the clockwise direction.

Here, all that is mentionedOf the sampling points, a certain sampling point is taken as a regression analysis point, and the spatial position coordinate of the regression analysis point is (x) _i ，y _i ) Determining the equivalent radius R of the ellipse search area according to the actual data condition (the space position coordinates of the area corresponding to each sampling point) of the research area _i Value range of [ R ] _min ,R _max ](suppose R _i ＝(R ₁ ，R ₂ ，…，R _n ) The interval between each equivalent radius can be equal or unequal, the step length (interval) can be set according to the actual situation, and the equivalent radius R is set _i The value range is split according to the set step length to determine a plurality of equivalent radiuses R ₁ ，R ₂ ，…，R _n 。

Illustratively, the range of the compression coefficient alpha is (0, 1]; the range of the azimuth angle theta is [0,180 DEG ] due to the symmetry of the ellipse; a series of parameter values can be set at equal intervals, and the setting of the intervals is determined by self-defining according to the actual data condition and the operation efficiency; for example, at intervals of 0.1 for a compression factor α, i.e., α =0.1,0.2,0.3, \ 8230;, 1; the interval of azimuth angle θ is 10 °, i.e., θ =0 °,10 °,20 °,30 °, \ 8230;, 170 °.

In step S102, the step of constructing a plurality of ellipse search regions having the regression analysis point as the center of the ellipse based on the predetermined ellipse search parameter includes:

s1021, acquiring a plurality of equivalent radiuses, a plurality of compression coefficients and a plurality of azimuth angles in the ellipse parameters according to the predetermined ellipse search parameters;

s1022, in the equivalent radiuses, taking the initial equivalent radius as the current equivalent radius, and constructing an ellipse search area corresponding to each azimuth angle of the compression coefficient under the current equivalent radius by taking the regression analysis point as the ellipse center for each compression coefficient;

s1023, determining whether a next equivalent radius exists;

s1024, if yes, updating the next equivalent radius to the current equivalent radius, and continuously constructing a plurality of ellipse searching areas taking the next equivalent radius as the current equivalent radius;

and S1025, if not, obtaining a plurality of ellipse search areas corresponding to each equivalent radius in all equivalent radii taking the regression analysis point as the center of the ellipse.

Illustratively, a plurality of equivalent radii, a plurality of compression coefficients and a plurality of azimuth angles in the ellipse parameters are obtained in step S206, and an initial equivalent radius R is selected ₁ (e.g., R at equal-spaced sampling times ₁ Is the sampling interval of the sampling point, R when sampling with unequal distances ₁ Minimum sample point spacing), the initial compression coefficient α = b/a =0.1, and the semi-major axis of the ellipse search area is calculated as

The minor semi-axis of the elliptical search area is b = α a, and the azimuth angle θ is set to 0 °; at the initial equivalent radius R ₁ And under the condition of (3), transforming the azimuth angle theta and the compression coefficient alpha to obtain a plurality of ellipse search areas under the equivalent radius.

For example, please refer to fig. 9, fig. 9 is an equivalent radius R provided by an embodiment of the present application ₁ -compression factor α ₁ For determining the equivalent radius size R, as shown in fig. 9 ₁ And a compression factor alpha ₁ The rotation angle (azimuth angle) is changed with the center sample point as the regression analysis point and the equivalent radius R ₁ And a compression factor alpha ₁ Varying the azimuth angle by θ =0 °,10 °,20 °,30 °, \8230, 170 °, generating a plurality of elliptical search areas of different azimuth angles under the condition. The compression factor can then also be varied, i.e. at the equivalent radius size R ₁ And a compression factor alpha ₂ The rotation angle (azimuth angle) is transformed under the condition, a plurality of ellipse search areas with different azimuth angles under the condition are generated until all compression coefficients are traversed, and the equivalent radius R is obtained ₁ The ellipse search area corresponding to each azimuth angle under each compression coefficient; finally, the equivalent radius is transformed to obtain the equivalent radius R ₂ Each orientation under each compressibility ofAnd searching the ellipse search areas corresponding to the angles until all the equivalent radiuses are traversed to obtain a plurality of ellipse search areas corresponding to each equivalent radius in all the equivalent radiuses with the regression analysis point as the center of the ellipse.

For example, if the sampling points are geochemical data, after detailed investigation and experimental evaluation in a region, considering that an ellipse search area with an equivalent radius of 500m covers less sampling points and has no statistical significance, setting the value range of the equivalent radius to [1000, 10000], namely the initial equivalent radius is 1000m, the maximum equivalent radius is 10000m which covers the maximum circle radius of all the sampling points, and the step length (interval) is 500m; intervals with a compressibility factor α of 0.1, i.e., α =0.1,0.2,0.3, \ 8230;, 1; and obtaining an elliptical search area corresponding to each azimuth angle of each compression coefficient under each equivalent radius of a regression analysis point when each sampling point is obtained as the regression analysis point by taking 10 degrees as the interval of the azimuth angles theta, namely theta =0 degrees, 10 degrees, 20 degrees, 30 degrees, 8230degrees and 170 degrees.

S103, determining a target ellipse searching region of the regression analysis point in a plurality of ellipse searching regions of the regression analysis point according to a space weighting U statistical method;

it should be noted that the step of determining the target elliptical search area of the regression analysis point from among the plurality of elliptical search areas of the regression analysis point includes:

s1031, in the multiple ellipse search areas of the regression analysis point, determining a spatial weighting U statistic corresponding to each ellipse search area in the multiple ellipse search areas corresponding to the equivalent radius aiming at each equivalent radius in the ellipse search parameters;

it should be noted that the spatial weighting U statistic corresponding to each elliptical search area is determined by the following steps:

1) Acquiring the spatial position coordinates of a region corresponding to each sampling point in all the sampling points in the oval search area and the concentration of a preset mineralizing element aiming at each oval search area;

2) Determining a long half shaft and a short half shaft of the ellipse search area according to the ellipse parameters corresponding to the ellipse search area;

in this step, for each elliptical search area, all sampling points in the elliptical search area are determined by the following formula:

wherein, a is a major semi-axis of the ellipse searching region, and b is a minor semi-axis of the ellipse searching region; searching for region B from each ellipse _x Parameters of (R, alpha, theta) by equivalent radius

And the compression factor α = b/a can be determined

And b = α a; and as long as the spatial position coordinates of the corresponding areas of the sampling points meet the formula, the sampling points are in the elliptical search area, so that all the sampling points in the elliptical search area are determined, and the spatial position coordinates of the corresponding areas of each sampling point and the concentration of the preset mineralizing elements are obtained.

3) Determining the target weight from each sampling point to the regression analysis point according to the major axis and the minor axis of the ellipse search area and the spatial position coordinates of the area corresponding to each sampling point in all the sampling points in the ellipse search area;

it should be noted that, the step of determining the target weight from each sampling point to the regression analysis point according to the major-axis and the minor-axis of the ellipse search area and the spatial position coordinates of the area corresponding to each sampling point in all the sampling points in the ellipse search area includes:

(1) Determining a weight parameter corresponding to each sampling point according to the major half axis and the minor half axis of the ellipse searching area and the spatial position coordinate of the area corresponding to each sampling point in all the sampling points in the ellipse searching area;

(2) Determining whether the weight parameter corresponding to each sampling point is in a range which is larger than zero and smaller than or equal to one or not for each sampling point;

(3) If so, squaring the weight parameter corresponding to the sampling point to obtain a first weight parameter, and squaring the difference between the first weight parameter and the first weight parameter to obtain the target weight from the sampling point to the regression analysis point;

(4) And if not, determining zero as the target weight from the sampling point to the regression analysis point.

4) And determining the spatial weighting U statistic corresponding to the elliptical search area according to the concentration of the predetermined ore-forming elements in the area corresponding to each sampling point and the target weight corresponding to each sampling point.

In this step, the formula for constructing the spatial weighting U statistic is as follows:

wherein z is _j For each sampling point in the elliptical search area, corresponding sampling value, such as geochemical element concentration value (concentration of predetermined mineralizing element), n is the number of sampling points in the elliptical search area, z ₀ A predetermined demarcation threshold for background and anomaly, z ₀ Typically taking a global or local mean, w _j And (5) a weight kernel function, namely the weight of the target determined in the step (4), wherein the weight value is related to the distance from the sampling point to the regression analysis point. Here, the abnormal state refers to a phenomenon that the content of some elements in the geologic body or natural substances in some regions is obviously deviated from the normal content or the chemical properties are obviously changed, and the phenomenon that the elements belong to the normal content is called as the background. If the sampling point in the ellipse searching area is divided into an abnormal part and a background part, a demarcation threshold value z ₀ The concentration mean value of the preset ore-forming elements in the area corresponding to all sampling points can be set; the demarcation threshold z is used if the anomaly and background are only divided in a local area ₀ Can be set to a local rangeThe concentration average value of the enclosed predetermined mineral forming elements. The programming of the model implementation can be selected from the two as desired.

When w is defined as _j When =1/n, i.e.

The same formula as for the spatial weighted U statistic. Considering the selection of the optimal ellipse search area of the spatial weighting U statistic method and the uneven distribution characteristics of the example geochemical data, the weight kernel function is improved based on the quadratic function in the truncated function, and the formula of the weight kernel function is as follows:

wherein d is _ij Here representing only one weight parameter, d _ij Is converted from an elliptical formula, and d _ij Is constantly not less than 0. When the sampling points are located in the ellipse search area and on the ellipse, d _ij Has a value range of (0, 1)]The weight value range of the improved weight kernel function is (0, 1)](ii) a If the sampling point is outside the ellipse search area, d _ij If the space weight value is larger than 1, the distance threshold characteristic of a truncation function method is adopted, and the space weight value is 0. The shape of the projection of the weight kernel function on the two-dimensional plane is an ellipse, as shown in fig. 9, the weight values on the ellipse boundary in the shadow are the same, and the distances between the sampling points on the ellipse boundary and the regression analysis points are different, and the distribution of the sampling points in the ellipse search region has directionality as viewed by the weight values.

S1032, in the space weighting U statistic corresponding to each ellipse searching area under the equivalent radius, determining the ellipse searching area with the largest absolute value of the space weighting U statistic as the optimal ellipse searching area of the equivalent radius;

and S1033, sorting the absolute values of the spatial weighting U statistics corresponding to the optimal ellipse search area of each equivalent radius according to each equivalent radius in the ellipse search parameters, and determining the optimal ellipse search area corresponding to the spatial weighting U statistics with the maximum absolute value as the target ellipse search area of the regression analysis point.

In the step, in the sampling points of each ellipse searching area, the sampling points are classified according to the positive and negative of the spatial weighting U statistic (taking geochemical data as an example, the spatial weighting U statistic is positive and is far more than 0, which indicates that the abnormal data of the sampling points in the ellipse searching area is more than background data, the larger the value is, the larger the proportion of the abnormal data is), according to the formula

At an equivalent radius R ₁ Selecting | U | N _ Y in a plurality of ellipse search areas _max The ellipse search area of (2) is the ellipse parameters (compression factor and azimuth) of the optimal ellipse search area at the equivalent radius. Transforming equivalent radius R _i Obtaining the ellipse parameters of the optimal ellipse search area under each equivalent radius

As shown in FIG. 8, all the spatially weighted U statistics of a sample point are sorted by absolute value, taking geochemical sampled data as an example, if | U _A | _max ＞|UB| _max At the sampling point

Therefore, the equivalent radius corresponding to the spatial weighting U statistic with the maximum absolute value is the optimal equivalent radius R ^* ＝R _A Meanwhile, it is explained that the sampling points in the ellipse search area using the sampling points as regression analysis points mostly belong to an abnormal group.

The spatial weighting U statistic reflects the relative difference between the concentration mean value of the predetermined ore-forming elements of the sampling points in the elliptic search area and a preset demarcation threshold value, and the classification of the abnormity and the background is to extract the optimal spatial weighting U statistic of an abnormal sampling point combination and a background sampling point combination in the elliptic search area of the regression analysis point; for example, at an equivalent radius R ₁ Down-converting the compression coefficient and azimuth angle, if the spatial weighting U statistic is positive and far larger than zero, indicating that the current value isMost of sampling points in an elliptical search area formed by the search parameters come from abnormity; on the contrary, if the spatial weighting U statistic is negative and far smaller than zero, it indicates that most of the sampling points in the elliptical search area formed by the current search parameter are from the background; the method is characterized in that the absolute value of the spatial weighting U statistic represents the quality of the classification effect of the sampling points in the formed elliptic search area, along with the change of the compression coefficient and the azimuth angle, when the absolute value | U | of the spatial weighting U statistic is maximum, the fact that the classification effect determined by the elliptic parameters corresponding to the current elliptic search area is optimal is shown, the spatial correlation of all sampling points in the elliptic search area is strongest, the vast majority of the sampling points in the elliptic search area are distributed in the same mode, and the compression coefficient and the azimuth angle at the moment are the optimal elliptic parameters under the equivalent radius.

The space weighting U statistic corresponding to the optimal ellipse searching area of each equivalent radius in each equivalent radius is determined according to the ordering of the equivalent radius, the optimal ellipse searching area corresponding to the space weighting U statistic with the maximum absolute value is determined as a target ellipse searching area, and the target ellipse searching area is output

Parameter R of ^* 、α ^* And theta ^* 。

S104, acquiring all sampling points included in the target ellipse searching region;

in this step, all sampling points included in the target ellipse search region are obtained by the following formula:

and when the space position coordinate of the area corresponding to the sampling point meets the formula, the sampling point is indicated to be in the target ellipse search area, so that all the sampling points in the target ellipse search area are obtained.

S105, determining a plurality of regression coefficients corresponding to the regression analysis point in the target ellipse search area according to the sampling point information of the area corresponding to each sampling point in the target ellipse search area and the predetermined target weight from each sampling point to the regression analysis point;

in this step, the sampling points obtained in step S104 are input into a traditional geographic weighted regression model for parameter estimation and model evaluation, and the ellipse parameters, the spatial weighted U statistical values, the regression coefficients, the residuals and the local goodness of fit of the target ellipse search area of each regression analysis point and the goodness of fit of the entire model are output. The traditional geo-weighted regression model is as follows:

wherein, y _i Is an observed value (concentration) of a dependent variable (predetermined mineralizing element) at a position i; x is the number of _ik Is the observed value (concentration) of the kth independent variable (predetermined geological activity sensitive element) at position i; (u) _i ,v _i ) The spatial position coordinates of the regression analysis point i; beta is a ₀ (u _i ,v _i ) Is an intercept term; beta is a _k (u _i ,v _i ) K = (1, 2, \8230;, p) is the kth regression coefficient at position i; epsilon _i Is an error term, follows a normal distribution with mean 0 and variance σ, and is independently identically distributed.

It should be noted that, the step of determining, according to the sampling point information of the area corresponding to each sampling point in the target ellipse search region and the predetermined target weight from each sampling point to the regression analysis point, a plurality of regression coefficients corresponding to the regression analysis point in the target ellipse search region includes:

s1051, determining a column vector composed of the concentrations of the predetermined mineralizing elements in the area corresponding to each sampling point in the target ellipse search area as a dependent variable column vector of a geographical weighted regression model according to the concentrations of the predetermined mineralizing elements in the sampling point information of the area corresponding to each sampling point;

s1052, determining a matrix formed by the concentrations of at least one preset geological activity sensitive element in the area corresponding to each sampling point in the target ellipse search area as an independent variable matrix of a geographical weighted regression model according to the concentration of at least one preset geological activity sensitive element in the sample point information of the area corresponding to each sampling point;

s1053, according to the predetermined target weight from each sampling point in the target ellipse search area to the regression analysis point, determining a diagonal matrix formed by the target weights from each sampling point to the regression analysis point as a space weight matrix of a geographical weighting regression model;

s1054, multiplying the transpose matrix of the independent variable matrix, the space weight matrix and the independent variable matrix to obtain a first matrix;

s1055, multiplying the inverse matrix of the first matrix by the transpose matrix of the independent variable matrix to obtain a second matrix;

s1056, multiplying the second matrix with the space weight matrix to obtain a third matrix;

s1057, multiplying the third matrix with the dependent variable column vector to obtain a regression coefficient column vector;

s1058, determining a plurality of regression coefficients corresponding to the regression analysis point of the target ellipse search area according to the regression coefficient column vector.

In the step, a weighted linear least square method is respectively adopted for solving a geographic weighted regression model for each sampling point i, and the regression parameter estimation is shown in the following formula:

wherein,

is y _i Estimate of (a), w _ij And a weight check function value (target weight) representing the weight value from the jth sampling point to the regression analysis point i around the regression analysis point. By selecting a compoundCalculating weight kernel function value w of sampling point by proper kernel function _ij Taking a quadratic function as an example, the same modified quadratic function as the spatial weighting U statistic is used as the kernel function. The matrix for the parameter solution equation can be simplified as the following formula:

wherein,

is a regression coefficient estimated value; x is an independent variable matrix, the first column of which takes the value of 1 to estimate the intercept term beta ₀ (u _i ，v _i ) (ii) a y is a dependent variable column vector; w (u) _i ，v _i ) A spatial weight matrix (n × n matrix) for the ith regression analysis point, which is also a diagonal matrix with diagonal element values from each sampling point to the regression analysis point (u) _i ，v _i ) Spatial weight value w of _ij 。

Here, a transposed matrix X of an argument matrix X is formed ^T Spatial weight matrix W (u) _i ，v _i ) Multiplying with an independent variable matrix X to obtain a first matrix X ^T W(u _i ，v _i ) X; the inverse matrix (X) of the first matrix ^T W(u _i ，v _i )X) ^-1 Transposed matrix X with argument matrix ^T Multiplying to obtain a second matrix (X) ^T W(u _i ，v _i )X) ^-1 X ^T (ii) a Second matrix (X) ^T W(u _i ，v _i )X) ^-1 X ^T And the spatial weight matrix W (u) _i ，v _i ) Multiplying to obtain a third matrix (X) ^T W(u _i ，v _i )X) ^-1 X ^T W(u _i ，v _i ) (ii) a The third matrix (X) ^T W(u _i ，v _i )X) ^-1 X ^T W(u _i ，v _i ) Multiplying the column vector y of the dependent variable to obtain a column vector of a regression coefficient

Column vector based on regression coefficients

Determining a plurality of regression coefficients beta corresponding to regression analysis points of a target ellipse search region ₀ (u _i ，v _i )，β ₁ (u _i ，v _i )，…，β _p (u _i ，v _i )。

And S106, creating a geographical weighting regression model corresponding to the target ellipse searching region with the regression analysis point as the center of the ellipse according to the regression coefficient, and analyzing the influence degree of the concentration of each preset geological activity sensitive element in the region corresponding to each sampling point in the target ellipse searching region with the regression analysis point as the center of the ellipse on the corresponding target mineralization element.

In this step, the regression coefficient β obtained in the above step S105 is used _k (u _i ，v _i ) K = (0, 2, \8230;, p) create a geographically weighted regression model for the target ellipse search region centered on the regression analysis point, as follows:

the dependent variable estimated value of the point i can be obtained according to the geographical weighted regression model

The formula is as follows:

wherein, X _i The ith row vector of the X matrix is represented.

And traversing all sampling points in the target research area to obtain a target ellipse searching area and a fitting equation of each sampling point, and obtaining ellipse parameters, local regression coefficients and residual errors of the target ellipse searching area corresponding to each sampling point so as to quantitatively describe the anisotropy and the spatial non-stationarity characteristics of the geochemical data.

It should be noted that, in terms of model evaluation, the goodness of fit R of the traditional geo-weighted regression model is used ² Goodness of fit R as an evaluation method of model fitting Effect ² E (0, 1), the degree to which the independent variable can account for the dependent variable, the formula is as follows:

wherein,

representing the dependent variable fit value, Y _i The value of the observed value is represented,

represents the mean of the observed values. R is ² The closer the value is to 1, the higher the firmness of the independent variable to the dependent variable is, the better the fitting degree of the regression equation is, and the more credible the result is.

Here, the embodiment is a general model improved for a traditional geographic weighted regression model, and regression analysis can be performed in different scenarios to achieve quantitative description of anisotropic features and non-stationary features of geospatial data.

For example, referring to fig. 10, fig. 10 is a flowchart of the work flow of the geographic weighted regression model based on the spatial weighted U statistic improvement provided by the embodiment of the present application, as shown in fig. 10, the steps include: (1) Input all data (x) ₁ ，x ₂ ，…，x _n ) (ii) a (2) Selecting an analysis point x ₁ (ii) a (3) Determining the sub-sample data group of the analysis point by a spatial weighting U statistical method; (4) Computing weights w for subsample data by a quadratic kernel function _ij (ii) a (5) Estimating regression parameters of the analysis points by using a weighted least square method, and determining whether j is equal to n; if n is equal to n, ending, if not equal to n, adding one to j and entering the step (3). The step (3) herein includes: is provided withSetting an initial equivalent radius R ₁ Azimuth θ =0 °,10 °,20 °,30 °, \8230, 170 °; the compression coefficient alpha =0.1,0.2,0.3, \8230, 1; according to

Alpha = b/a calculating ellipse length semi-axis

b = α a; transforming the azimuth angle theta, compressing the coefficient alpha and obtaining a series of elliptical windows under the equivalent radius; according to the formula:

acquiring sub-sample data in each elliptical window; calculating the spatial weighting U statistic of the sub-sample data in each elliptical window; to find out

Transformation of R _i ＝(R ₂ ，…，R _m ) Obtaining the optimal elliptical window under each equivalent radius

Sorting by equivalent radius size, selecting the largest

Output parameter U ^* (B _x (R, α, θ)) in the elliptical window.

Referring to fig. 11, fig. 11 is a workflow for non-stationarity analysis of geochemical data space of a mine according to an embodiment of the present application, and as shown in fig. 11, the steps include: the method comprises the steps of obtaining geochemical data and establishing an ore deposit model by inquiring literature data and data collection or/and field investigation sampling analysis, determining independent variables and dependent variables, and performing model evaluation comparison on a GWR model and a traditional GWR model by improving the GWR model based on a spatial weighting U statistical method. Step of improving GWR model by using space weighting U statistic methodThe method comprises the following steps: selecting regression point, determining a series of variable ellipse windows with regression point as ellipse center, screening subsample group, calculating U value, determining | U | _max And obtaining an optimal local elliptical window and a regression equation, traversing the whole region and obtaining the regression equation of each point. The steps of the conventional GWR model include: selecting a regression point, determining a series of changed circular windows with the regression point as the center of a circle, screening a sub-sample group, moving to the whole area to obtain a plurality of groups of regression equations, calculating whether the AIC value is minimum, and if not, re-screening sub-sample data through the series of changed circular windows; if so, acquiring the optimal bandwidth b, traversing the whole region under the condition of the optimal bandwidth b, and acquiring the regression equation of each point.

Illustratively, the improved GWR and the traditional GWR models are respectively applied to the space non-stationarity analysis of geochemical sampling data of 1:5 ten thousand water system sediments in a mining area, the copper-gold symbiosis relation is quantitatively researched by a statistical analysis method such as the traditional GWR model in the prior art, but a general linear regression method can only determine the general trend, scattering modes in different modes still exist in a local area, the traditional GWR model does not consider the anisotropy characteristic of a geological process, and the improved GWR model has defects in classification screening of sampling point geochemical data background and abnormal distribution. In the geographical weighted regression model improved based on the spatial weighted U statistic method provided by the embodiment, the ellipse search window and the spatial weighted U statistic value in the spatial weighted U statistic method are used for selecting the sample, so that the geochemical data in the local range can be better classified and screened, and therefore, the improved GWR model is more reasonable in application in the copper-gold symbiotic relationship research and has deeper geological significance. The two models are compared and evaluated through regression coefficients and evaluation factors, the geochemical distribution of the mineralization elements copper and gold in a mining area is researched, the spatial variation characteristics of the geochemical distribution are analyzed, the background abnormal separation effect realized by the spatial weighting U statistic in the GWR model is discussed and improved, the local anisotropy characteristics of the regression result of the geochemical data model under the influence of the mineralization are discussed, and the understanding of the local mineralization mechanism in the mining area is deepened.

Illustratively, after logarithmic preprocessing is performed on the concentration values of copper (%) and gold (g/t) elements of 3217 geochemical sampling points of a target research area (such as a mining area), regression analysis is respectively performed as a dependent variable (copper Cu as a predetermined mineral forming element) and an independent variable (gold Au as a predetermined geological activity sensitive element) of a geographic weighted regression model, and a one-factor geographic weighted regression model is constructed as follows:

ln(Cu)＝β ₀ (u _i ，v _i )+β ₁ (u _i ，v _i )ln(Au)+ε _i ，i＝1，2，…，n；

wherein (u) _i ，v _i ) The spatial position coordinates (such as longitude and latitude coordinates) of the area corresponding to the ith sampling point; beta is a _k (u _i ，v _i ) Is the kth regression coefficient for the ith sample point (this time an argument, k =0, 1), which is a function of geographic location; error term epsilon _i Obey a normal distribution with mean 0 and variance σ, and are independently identically distributed, i.e.: epsilon _i ～N(0，σ ² )，Cov(ε _i ，ε _j )＝0(i≠j)。

The weight kernel function is an improved quadratic function, the improved geographical weighted regression model GWR is evaluated and analyzed, the sampling point in the local elliptic search area of the improved GWR model is uncertain, and the GWR model is composed of | U | _max Determining the range of sampling points in the local ellipse searching area by the value; on the evaluation of the overall effect of the model, the GWR model is improved to pass through the coefficient of block R ² Evaluating the fitting effect of the model, and obtaining an improved model R through experimental data ² The value is larger than 0.75, the effect of improving the GWR model is good, sampling points are selected by the ellipse search area and the space weighting U statistic value in the space weighting U statistic method in the GWR model, and the purpose is to better classify and screen geochemical data in a local range, so that the model is more meaningful to improve when the model is applied to the copper-gold symbiotic relation.

Here, from the viewpoint of model prediction, the conventional type GWR model (taking fixed bandwidth as an example) and the improved GWR model are used for estimation of the dependent variable Cu element concentration and inverse distance weighted interpolation is performed. Referring to fig. 12, fig. 13 and fig. 14, fig. 12 is an interpolation graph of observed values of predetermined mineralizing element concentrations provided in an embodiment of the present application, fig. 13 is an interpolation graph of estimated values of predetermined mineralizing element concentrations based on a conventional GWR model provided in an embodiment of the present application, and fig. 14 is an interpolation graph of estimated values of predetermined mineralizing element concentrations of an improved GWR model based on spatial U statistics provided in an embodiment of the present application; as shown in FIG. 12, the interpolation result of the measured values of the Cu element concentration shows that deposit points are spatially located in the high-value concentration area of the Cu element concentration. The model visualization analysis results are shown in fig. 13 and 14, the estimation results of the traditional GWR model and the improved GWR model for the dependent variable Cu element concentration conform to the spatial distribution rule of the original measurement values, compared with the case that the high value and the low value of the improved GWR model for the dependent variable Cu estimation value are closer to the original Cu element concentration observation value, and the prediction results show that the improved GWR model is more effective in algorithm improvement.

Here, referring to fig. 15 and 16, and fig. 17 and 18, fig. 15 is a graph illustrating a regression coefficient β based on a conventional GWR model according to an embodiment of the present application ₀ Results interpolation plot, figure 16 is a graph of regression coefficients β for improved GWR models based on spatial U statistics as provided in the examples of the present application ₀ Result interpolation graph, fig. 17 is a graph of regression coefficient β based on the conventional GWR model provided in the example of the present application ₁ Results interpolation plot, figure 18 is a graph of the regression coefficient β of the improved GWR model based on the spatial U statistics as provided by an embodiment of the present application ₁ A result interpolation graph; in terms of model local regression coefficients, the regression coefficients may represent the degree of influence of the independent variable (Au) on the dependent variable (Cu), the higher the degree of influence of gold on the copper element near the deposit point, the higher the coefficient β ₀ (u _i ，v _i ) The spatial distribution of (A) is shown in FIGS. 15 and 16, and beta ₁ (u _i ，v _i ) As shown in fig. 17 and 18, the regression coefficients of the two models are estimated to have similar spatial distribution, with high values distributed in the region near the deposit point, but the improved GWR model regression coefficients are spatially distributed in more detail and highlight the characteristic of the spatial anisotropy that the effect of gold on copper element is expressed by fracture formation control.

Here, please refer to fig. 19 and fig. 20, where fig. 19 is a partial R based on a conventional GWR model provided by the embodiment of the present application and provided by the embodiment of the present application ² As a result, theInterpolation graph, fig. 20 is a partial R of GWR model improved based on spatial U statistics provided by an embodiment of the present application ² Result interpolation plot, in GWR model, the coefficient R can be determined ² Not only can the overall fitting effect of the model be evaluated, but also the fitting effect in each ellipse searching region can be evaluated, the fitting effect is changed in the whole research region, and the symbiotic combination of copper and gold in the mining process is not an absolute linear process, so that the local R is used ² The value can be expressed as the strength of the copper-gold symbiotic relation of the ore forming area and the coefficient R ² The higher the value, the stronger the copper-gold symbiotic relationship, the more favorable the mineralization. Local R of two models ² The interpolation results are shown in fig. 19 and 20, and it can be seen that the improved GWR model is better than the conventional GWR model in terms of overlapping of a high-value area of the coefficient of block, a fault intersection and an ore deposit mineralization area, because the GWR model not only separates the background and abnormal distribution of the predetermined mineralization elements in the area corresponding to the sampling point when the sampling point of the ellipse search area is selected, but also reflects the anisotropic characteristics of the determination of the target ellipse search area and the weight value of the kernel function, that is, the determination of the target ellipse search area is affected by the fracture structure. At R ² The overall spatial distribution mode is used for evaluating the strength of the copper-gold symbiosis relationship in the research area, and the GWR model is improved to be more advantageous.

For example, referring to fig. 21 and fig. 22, fig. 21 is a graph illustrating a space U statistic based improved GWR model multi-factor regression local R provided by an embodiment of the present application ² A result interpolation graph, fig. 22 is a graph of interpolation of multi-factor regression residual results for improving GWR model based on spatial U statistics provided in this embodiment of the present application, in order to further verify the co-occurrence relationship between cu and au and the evaluation factor (local coefficient of solution R) in the GWR model ² ) The relation of (2) and the discussion of the defects of a model residual error mode, a multi-factor GWR model is established for a main mineral element Cu and structural activity sensitive elements Au, as, sb and Hg in a target research area (mining area), and the formula is As follows:

ln(Cu)＝β ₀ (u _i ，v _i )+β ₁ (u _i ，v _i )ln(Au)+β ₂ (u _i ，v _i )ln(As)+β ₃ (u _i ，v _i )ln(Sb)+β ₄ (u _i ，v _i )ln(Hg)+ε _i ，(i＝1，2，…，n)；

improving local coefficient of block R of GWR model ² And residual results are shown in fig. 21 and 22. Under the multifactor regression model, R ² The spatial distribution of high values is relatively uniform, and the high values and the low values are distributed in a cross way without obvious rules; while in the one-factor regression model, R ² High values are obviously gathered in the porphyry deposit area and the fracture intersection, and the local R is verified to pass through laterally ² Shows the effectiveness of the strong and weak symbiotic relationship of Cu and Au, and local R ² The higher the value is, the stronger the copper-gold intergrowth relationship is, and the advantage of improving the GWR model to express anisotropic characteristics is more beneficial to describing the ore-forming control effect. The overall coefficient of the improved GWR in the multi-factor regression model is 0.79, and the fitting effect is improved compared with that of single-factor regression. From the size of the residual high value, after other associated elements in the mineralization process are added in the model to serve as new independent variables, the residual high value is relatively reduced from 0.823 to 0.673, and the model is correct in the aspect of adding the independent variable type on the estimation of the dependent variable copper element by combining the integral coefficient of the model; from the spatial distribution of the residual errors, the high-value area and the low-value area of the multi-factor regression model are distributed in a cross mode, the relative distribution is uniform, the residual error value of the area where the ore deposit is located is slightly higher than that of most areas of a research area (ore area), and the fact that the statistical reason is that the GWR model residual error mode is insufficient is verified.

Here, the improved GWR model is applied in an exemplary application for exploring geochemical data by means of a locally determined coefficient R ² And evaluation factors such as residual errors and the like evaluate model regression results, can be used for describing existence and strength degree of copper-gold symbiotic relationship, researching multi-factor regression results of tectonic activity sensitive elements on mineralizing elements, and can be used for conducting geological significance discussions such as anisotropic space quantitative analysis of geochemical elements, mineralizing mechanism and the like. Compared with the traditional GWR model, the improved GWR model is used for describing complex data by deeply improving multiple aspects such as weight kernel function (namely target weight), bandwidth optimal determination mode (namely determination of target ellipse search area), bandwidth parameter (namely ellipse parameter) setting and the likeWhen the spatial data change, all spatial non-stationarity characteristics of a traditional GWR model are reserved, anisotropic characteristics of data spatial distribution are developed, quantitative description of data spatial change is achieved, application scenarios of a spatial U statistical method are greatly expanded, the spatial U statistical method is innovatively used for determining the optimal local search elliptical area of a geographic weighting regression model, a new method is provided for analyzing the anisotropy of geographic spatial data, the applicability of the GWR model is improved, and the method can be applied to various data types and application requirements; in a geological application scene of space non-stationarity and anisotropy analysis of the control effect of the ore control elements on the ore forming process, the improved GWR model has obvious advantages, the classification effect of geochemical data background and abnormal distribution can be improved, the quantitative analysis and description of the copper-gold symbiotic relationship of a mining area and geological activities related to ore forming can be realized, and the spatial distribution of a complex geological process and products thereof can be objectively reflected.

The method for establishing the geographical weighted regression model for analyzing the correlation of the ore-forming elements, provided by the embodiment of the application, comprises the following steps: acquiring sampling point information of a region corresponding to a plurality of sampling points in a target research region; the sampling point information comprises spatial position coordinates of a region corresponding to the sampling point, the concentration of a preset mineralizing element and the concentration of at least one preset geological activity sensitive element; taking any one of the sampling points as a regression analysis point, and constructing a plurality of ellipse search areas taking the regression analysis point as an ellipse center according to predetermined ellipse search parameters; determining a target ellipse searching region of the regression analysis point in a plurality of ellipse searching regions of the regression analysis point according to a space weighting U statistical method; acquiring all sampling points included in the target ellipse searching area; determining a plurality of regression coefficients corresponding to the regression analysis point in the target ellipse search area according to sampling point information of a region corresponding to each sampling point in the target ellipse search area and a predetermined target weight from each sampling point to the regression analysis point; and creating a geographical weighted regression model corresponding to the target ellipse searching region with the regression analysis point as the ellipse center according to the regression coefficient, so as to be used for analyzing the influence degree of the concentration of each preset geological activity sensitive element in the region corresponding to each sampling point in the target ellipse searching region with the regression analysis point as the ellipse center on the corresponding target mineralization element.

Based on the same application concept, a device for creating a geographical weighted regression model for analyzing correlation of mineral forming elements, which corresponds to the method for creating a geographical weighted regression model for analyzing correlation of mineral forming elements provided in the foregoing embodiments, is also provided in the present embodiment.

Referring to fig. 23 and 24, fig. 23 is a first structural diagram of a geographic weighted regression model creation apparatus for analyzing correlation of mineral elements according to an embodiment of the present application, and fig. 24 is a second structural diagram of the geographic weighted regression model creation apparatus for analyzing correlation of mineral elements according to an embodiment of the present application. As shown in fig. 23, the creating means 2310 includes:

a first obtaining module 2311, configured to obtain sampling point information of an area corresponding to multiple sampling points in a target research area; the sampling point information comprises spatial position coordinates of a region corresponding to the sampling point, the concentration of a preset mineralization element and the concentration of at least one preset geological activity sensitive element;

a constructing module 2312, configured to construct, using any one of the multiple sampling points as a regression analysis point, multiple ellipse search regions using the regression analysis point as an ellipse center according to a predetermined ellipse search parameter;

a first determining module 2313, configured to determine a target ellipse search region of the regression analysis point in the plurality of ellipse search regions of the regression analysis point according to a spatial weighted U-statistic method;

a second obtaining module 2314, configured to obtain all sampling points included in the target ellipse search area;

a processing module 2315, configured to determine, according to sample point information of a region corresponding to each sampling point in the target ellipse search region and a predetermined target weight from each sampling point to the regression analysis point, a plurality of regression coefficients corresponding to the regression analysis point in the target ellipse search region;

a creating module 2316, configured to create, according to the regression coefficient, a geographic weighted regression model corresponding to the target ellipse search region with the regression analysis point as an ellipse center, so as to analyze the influence degree of the concentration of each predetermined geological activity sensitive element in the area corresponding to each sampling point in the target ellipse search region with the regression analysis point as the ellipse center on the corresponding target mineralization element.

Optionally, as shown in fig. 24, the creating apparatus 2310 further includes a second determining module 2317, where the second determining module 2317 is configured to:

determining a long half shaft range and a short half shaft range of an ellipse search area according to the space position coordinates of the area corresponding to each sampling point;

splitting the value range of the equivalent radius according to the step length and a preset step length to determine a plurality of equivalent radii;

splitting the value range of the compression coefficient according to the preset compression coefficient interval according to the preset value range of the compression coefficient and the preset compression coefficient interval, and determining a plurality of compression coefficients;

splitting the value range of the azimuth angle according to a preset azimuth interval according to the value range of a preset azimuth angle and a preset azimuth interval, and determining a plurality of azimuth angles;

Optionally, when the constructing module 2312 is configured to construct, according to the predetermined ellipse search parameter, a plurality of ellipse search areas with the regression analysis point as an ellipse center, the constructing module 2312 is specifically configured to:

determining whether there is a next equivalent radius;

Optionally, when the first determining module 2313 is configured to determine the target elliptical search area of the regression analysis point from the multiple elliptical search areas of the regression analysis point, the first determining module 2313 is specifically configured to:

in the space weighting U statistic corresponding to each ellipse searching area under the equivalent radius, determining the ellipse searching area with the maximum absolute value of the space weighting U statistic as the optimal ellipse searching area of the equivalent radius;

and sequencing the absolute values of the spatial weighting U statistics corresponding to the optimal ellipse searching area of each equivalent radius according to each equivalent radius in the ellipse searching parameters, and determining the optimal ellipse searching area corresponding to the spatial weighting U statistics with the maximum absolute value as the target ellipse searching area of the regression analysis point.

Optionally, when the first determining module 2313 is configured to determine the spatial weighting U statistic corresponding to each elliptical search area, the first determining module 2313 is specifically configured to:

and determining the spatial weighting U statistic corresponding to the ellipse search area according to the concentration of the predetermined ore-forming elements in the area corresponding to each sampling point and the target weight corresponding to each sampling point.

Optionally, when the first determining module 2313 is configured to determine the target weight from each sampling point to the regression analysis point according to the semi-major axis and the semi-minor axis of the ellipse search area and the spatial position coordinate of the area corresponding to each sampling point in all sampling points in the ellipse search area, the first determining module 2313 is specifically configured to:

determining a weight parameter corresponding to each sampling point according to the major half axis and the minor half axis of the ellipse searching region and the spatial position coordinate of the area corresponding to each sampling point in all the sampling points in the ellipse searching region;

determining whether the weight parameter corresponding to each sampling point is in a range which is larger than zero and smaller than or equal to one or not for each sampling point;

and if not, determining zero as the target weight from the sampling point to the regression analysis point.

Optionally, when the processing module 2315 is configured to determine, according to the sampling point information of the area corresponding to each sampling point in the target ellipse search area and the predetermined target weight from each sampling point to the regression analysis point, a plurality of regression coefficients corresponding to the regression analysis point in the target ellipse search area, the processing module 2315 is specifically configured to:

determining a matrix formed by the concentrations of at least one predetermined geological activity sensitive element in the area corresponding to each sampling point in the target ellipse search area as an independent variable matrix of a geographic weighted regression model according to the concentration of at least one predetermined geological activity sensitive element in the sample point information of the area corresponding to each sampling point;

determining a diagonal matrix formed by the target weights from each sampling point to the regression analysis point as a space weight matrix of a geographical weighted regression model according to the predetermined target weights from each sampling point to the regression analysis point in the target ellipse search area;

multiplying the transposed matrix of the independent variable matrix, the space weight matrix and the independent variable matrix to obtain a first matrix;

The geographical weighted regression model creation device for analyzing the correlation of the ore-forming elements provided by the embodiment of the application comprises: the first acquisition module is used for acquiring sampling point information of a region corresponding to a plurality of sampling points in a target research region; the sampling point information comprises spatial position coordinates of a region corresponding to the sampling point, the concentration of a preset mineralization element and the concentration of at least one preset geological activity sensitive element; the construction module is used for constructing a plurality of ellipse search areas which take any one of the plurality of sampling points as a regression analysis point and take the regression analysis point as an ellipse center according to predetermined ellipse search parameters; the first determining module is used for determining a target ellipse searching region of the regression analysis point in a plurality of ellipse searching regions of the regression analysis point according to a space weighting U statistical method; the second acquisition module is used for acquiring all sampling points in the target ellipse search area; the processing module is used for determining a plurality of regression coefficients corresponding to the regression analysis point in the target ellipse search area according to sampling point information of a region corresponding to each sampling point in the target ellipse search area and a predetermined target weight from each sampling point to the regression analysis point; and the creating module is used for creating a geographical weighted regression model corresponding to the target ellipse searching region with the regression analysis point as the ellipse center according to the regression coefficient, and is used for analyzing the influence degree of the concentration of each preset geological activity sensitive element in the region corresponding to each sampling point in the target ellipse searching region with the regression analysis point as the ellipse center on the corresponding target mineralization element.

Referring to fig. 25, fig. 25 is a schematic structural diagram of an electronic device according to an embodiment of the present disclosure. As shown in fig. 25, the electronic device 2500 includes a processor 2510, a memory 2520, and a bus 2530.

The memory 2520 stores machine readable instructions executable by the processor 2510, when the electronic device 2500 runs, the processor 2510 and the memory 2520 communicate through the bus 2530, and when the processor 2510 executes the machine readable instructions, the steps of the method for creating a geographical weighted regression model for analyzing correlation between mineralogical elements in the method embodiments shown in fig. 1 and fig. 2 may be executed, and specific implementation manners may refer to the method embodiments, which are not described herein again.

An embodiment of the present application further provides a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the step of the method for creating a geographic weighted regression model for analyzing correlations between mineralization elements in the method embodiments shown in fig. 1 and fig. 2 may be executed.

It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.

In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other ways. The above-described apparatus embodiments are merely illustrative, and for example, the division of the units into only one type of logical function may be implemented in other ways, and for example, multiple units or components may be combined or integrated into another system, or some features may be omitted, or not implemented. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection of devices or units through some communication interfaces, and may be in an electrical, mechanical or other form.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit.

The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a non-volatile computer-readable storage medium executable by a processor. Based on such understanding, the technical solution of the present application or portions thereof that substantially contribute to the prior art may be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.

Finally, it should be noted that: the above-mentioned embodiments are only specific embodiments of the present application, and are used for illustrating the technical solutions of the present application, but not limiting the same, and the scope of the present application is not limited thereto, and although the present application is described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: those skilled in the art can still make modifications or changes to the embodiments described in the foregoing embodiments, or make equivalent substitutions for some features, within the technical scope of the present disclosure; such modifications, changes or substitutions do not depart from the spirit and scope of the exemplary embodiments of the present application, and are intended to be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A method for creating a geographically weighted regression model for analyzing correlation of mineralogical elements, the method comprising:

acquiring sampling point information of a region corresponding to a plurality of sampling points in a target research region; the sampling point information comprises spatial position coordinates of a region corresponding to the sampling point, the concentration of a preset mineralizing element and the concentration of at least one preset geological activity sensitive element;

taking any one of the plurality of sampling points as a regression analysis point, and constructing a plurality of ellipse search areas taking the regression analysis point as an ellipse center according to predetermined ellipse search parameters;

acquiring all sampling points included in the target ellipse searching area;

2. The method of creating according to claim 1, characterized by an ellipse search parameter determined by the steps of:

determining a value range of the equivalent radius of the ellipse search area according to the range of the major semi-axis and the range of the minor semi-axis of the ellipse search area;

3. The method for creating a new search space according to claim 1, wherein the step of constructing a plurality of elliptical search areas with the regression analysis point as a center of an ellipse according to predetermined elliptical search parameters comprises:

determining whether there is a next equivalent radius;

4. The method of creating as claimed in claim 1, wherein the step of determining the target elliptical search area of the regression analysis point among the plurality of elliptical search areas of the regression analysis point comprises:

5. The method of creating as claimed in claim 4 wherein the spatial weighting U statistic for each elliptical search area is determined by:

6. The creating method according to claim 5, wherein the step of determining the target weight from each sampling point to the regression analysis point according to the semi-major axis and the semi-minor axis of the elliptical search area and the spatial position coordinates of the area corresponding to each sampling point in all the sampling points in the elliptical search area comprises:

7. The creating method according to claim 1, wherein the step of determining a plurality of regression coefficients corresponding to the regression analysis point in the target ellipse search region according to the sampling point information of the area corresponding to each sampling point in the target ellipse search region and the predetermined target weight from each sampling point to the regression analysis point comprises:

8. A geoweighted regression model creation apparatus that analyzes correlation of mineralogical elements, the creation apparatus comprising:

the processing module is used for determining a plurality of regression coefficients corresponding to the regression analysis point in the target ellipse search area according to the sampling point information of the area corresponding to each sampling point in the target ellipse search area and the predetermined target weight from each sampling point to the regression analysis point;

9. An electronic device, comprising: a processor, a memory and a bus, the memory storing machine readable instructions executable by the processor, the processor and the memory communicating over the bus when the electronic device is run, the machine readable instructions when executed by the processor performing the steps of the method of creating a geographically weighted regression model for analysis of correlation of mineralogical elements as claimed in any one of claims 1 to 7.

10. A computer-readable storage medium, having stored thereon a computer program for performing, when being executed by a processor, the steps of the method for creating a geoweighted regression model for analyzing relevance of mineralogical elements according to any one of claims 1 to 7.