CN114493245A - Mountain torrent disaster easiness evaluation method based on GIS and integrated learning - Google Patents
Mountain torrent disaster easiness evaluation method based on GIS and integrated learning Download PDFInfo
- Publication number
- CN114493245A CN114493245A CN202210078561.8A CN202210078561A CN114493245A CN 114493245 A CN114493245 A CN 114493245A CN 202210078561 A CN202210078561 A CN 202210078561A CN 114493245 A CN114493245 A CN 114493245A
- Authority
- CN
- China
- Prior art keywords
- mountain torrent
- model
- easiness
- mountain
- algorithm
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000011156 evaluation Methods 0.000 title claims abstract description 43
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 56
- 238000000034 method Methods 0.000 claims abstract description 25
- 238000010801 machine learning Methods 0.000 claims abstract description 23
- 238000011160 research Methods 0.000 claims abstract description 12
- 230000010354 integration Effects 0.000 claims abstract description 7
- 238000012360 testing method Methods 0.000 claims description 24
- 238000012549 training Methods 0.000 claims description 18
- 238000013507 mapping Methods 0.000 claims description 12
- 238000007781 pre-processing Methods 0.000 claims description 9
- 230000003993 interaction Effects 0.000 claims description 7
- 238000012795 verification Methods 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 5
- 230000009471 action Effects 0.000 claims description 3
- 238000004140 cleaning Methods 0.000 claims description 3
- 230000008595 infiltration Effects 0.000 claims description 3
- 238000001764 infiltration Methods 0.000 claims description 3
- 230000007246 mechanism Effects 0.000 claims description 3
- 238000007619 statistical method Methods 0.000 claims description 3
- 238000007637 random forest analysis Methods 0.000 abstract description 21
- 238000013528 artificial neural network Methods 0.000 abstract description 12
- 238000011835 investigation Methods 0.000 abstract description 5
- 230000003044 adaptive effect Effects 0.000 abstract description 3
- 230000002265 prevention Effects 0.000 abstract description 3
- 238000004364 calculation method Methods 0.000 abstract description 2
- 238000004458 analytical method Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 230000006870 function Effects 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000035772 mutation Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007123 defense Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 230000007786 learning performance Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012502 risk assessment Methods 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/29—Geographical information databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/20—Ensemble learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A10/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE at coastal zones; at river basins
- Y02A10/40—Controlling or monitoring, e.g. of flood or hurricane; Forecasting, e.g. risk assessment or mapping
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Tourism & Hospitality (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- General Engineering & Computer Science (AREA)
- Educational Administration (AREA)
- Development Economics (AREA)
- General Health & Medical Sciences (AREA)
- General Business, Economics & Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Marketing (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Operations Research (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Remote Sensing (AREA)
- Medical Informatics (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Primary Health Care (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a mountain torrent disaster susceptibility evaluation method based on GIS and ensemble learning, which is characterized in that a Multivariate Adaptive Regression Spline (MARS) and a Random Forest (RF) algorithm are used as an ensemble learning basic model, an Artificial Neural Network (ANN) algorithm is used for integration, a plurality of small watershed vector units are used as research objects, and a plurality of mountain torrent characteristic factors are used as regression independent variables. The method takes the vector small watershed as a basic research unit, and can more accurately analyze the easiness degree of the small watershed area, so that the small watershed where the mountain torrents are easy to occur is accurately positioned. Meanwhile, the mountain torrent disaster investigation and evaluation result is combined with machine learning and an intelligent algorithm, so that mountain torrent disaster easiness evaluation and drawing are performed, the provided evaluation method is accurate in calculation, reliable in principle and easy and convenient to learn and operate, can be generally suitable for the requirement of provincial scale mountain torrent disaster easiness evaluation, and provides an auxiliary decision for prevention and control of mountain torrent disasters.
Description
Technical Field
The invention relates to the technical field of mountain torrent disaster easiness evaluation and mapping, in particular to a mountain torrent disaster easiness evaluation method based on GIS and ensemble learning.
Background
Risk assessment and mapping of mountain torrent easy-to-send areas are effective means for preventing mountain torrent disasters, various patents are used for assessing and mapping mountain torrent risk easy-to-send performance at present, and in past researches, grid is mainly used as an evaluation unit for analyzing mountain torrent easy-to-send performance, for example, a road mountain torrent vulnerability evaluation method based on GIS and machine learning disclosed in patent 202110274765.4;
in the analysis method of the easiness of the mountain torrent disaster, the traditional space analysis method is widely applied, and the traditional evaluation methods such as an AHP (advanced high performance analysis) analytic hierarchy process and a mutation theory evaluation method, for example, a mountain torrent disaster risk evaluation method early warning system based on a mutation theory disclosed in patent 201810117484.6, generally use expert knowledge to set index weight;
with the continuous development of machine learning and the proposal of geospatial intelligence (GeoAI), various machine learning methods have also been used in the field of big data assisted torrential flood modeling, such as a road torrential flood vulnerability evaluation method based on GIS and machine learning disclosed in patent 202110274765.4;
in addition, the effect of the single model can be improved through algorithm integration and model optimization, and further the machine learning model is used as a basic classifier, an integrated learning method is provided, and results of most learners show that the integrated learning has better effect on processing the nonlinear problem than the machine learning.
The invention provides a mountain torrent disaster easiness evaluation method based on GIS and integrated learning, which aims to effectively utilize vector small watershed data of mountain torrent disaster investigation evaluation results, accurately evaluate the mountain torrent easiness degree in each small watershed, study the performance of different machine learning in the aspect of mountain torrent evaluation and overcome the problem of poor effect of single machine learning in processing nonlinear problems.
Disclosure of Invention
The invention provides a mountain torrent disaster easiness evaluation method based on GIS and ensemble learning, which can effectively solve the problem that single machine learning is poor in effect on processing the nonlinear problem in the background technology.
In order to achieve the purpose, the invention provides the following technical scheme: a mountain torrent disaster proneness evaluation method based on GIS and ensemble learning is characterized in that a Multivariate Adaptive Regression Spline (MARS) and a Random Forest (RF) algorithm are used as an ensemble learning basic model, an Artificial Neural Network (ANN) algorithm is used for integration, a plurality of small watershed vector units are used as research objects, and a plurality of mountain torrent characteristic factors are used as regression independent variables;
the evaluation method specifically comprises the following steps:
firstly, preprocessing data;
secondly, establishing a model;
thirdly, the characteristic factors influence;
fourthly, verifying the performance of the algorithm;
and fifthly, drawing the mountain torrents easily.
According to the technical scheme, in the first step, the data preprocessing refers to preprocessing a small watershed data set, an evaluation index system is established according to the action mechanism of small watershed parameters on torrential flood disasters and according to an index selection principle, and indexes are considered from the three aspects of geometric characteristics (A1), hydrological characteristics (A2) and meteorological characteristics (A3).
According to the technical scheme, the geometric characteristics (A1) are selected from an average slope (A11), a weighted average slope (A12), a shape coefficient (A13), a longest confluence path length (A14), a longest confluence path slope (A15) and a longest confluence path slope 1085 (A16);
selecting roughness (A21), stable infiltration rate (A22) and unit line peak modulus (A23) with the frequency of 1% in 10 minutes from hydrological characteristics (A2);
selecting unit line data (A31) with the frequency of 1% in 10 minutes, rainfall (A32) with the frequency of 1% in 10 minutes, unit line data (A33) with the frequency of 1% in 30 minutes and rainfall (A24) with the frequency of 1% in 30 minutes from meteorological characteristics (A3);
on the basis of data cleaning, standardization and downsampling, a small watershed data set is trained, and 70% of samples are randomly selected as a training set and 30% of samples are selected as a test set to be used as model input.
According to the technical scheme, in the second step, the model building means that the selected multiple characteristic factors are used, two machine learning algorithms and an integration algorithm are adopted to evaluate the easiness of the small watershed torrential flood disasters, and the method specifically comprises the following steps:
firstly, using a training set to initially train an algorithm;
subsequently, in order to enable the algorithm to achieve the best performance, the optimization algorithm, the preliminary result and the test set data are adopted to optimize the parameters;
finally, the indexes are evaluated by using the test set data calculation algorithm, and the lifting algorithms are compared in detail.
According to the technical scheme, in order to overcome the unbalanced problem in mountain torrent proneness modeling and ensure the same proportion in the machine learning training model, after a plurality of adding models are randomly selected in other small watersheds without mountain torrents, training sets and test sets are respectively and randomly divided on mountain torrent and non-mountain torrent data sets according to the proportion of 7: 3, and the training sets and the test sets of a plurality of data are obtained.
According to the technical scheme, in the second step, the model building comprises the steps of completing the primary building of the MARS model by using the R language, completing the primary building of the RF model by using the R language, and completing the primary building of the ANN model by using the R language.
According to the technical scheme, in the third step, the influence of the characteristic factors refers to the contribution degree of the characteristic factors to the easiness of mountain torrent disasters;
the influence of single contribution of the characteristic factors and the influence of double-factor interaction on the mountain torrent disaster easiness are obtained according to results of the MARS model and the RF model, and the influence of factor importance in the RF model and the influence of multi-factor interaction in the MARS model on the mountain torrent disaster easiness are mainly checked.
According to the technical scheme, in the fourth step, the verification algorithm performance refers to model verification and comparison, the algorithm precision is verified mainly by running test set data, in the test algorithm performance part, the precision is evaluated by calculating algorithm evaluation indexes including precision, recall rate, precision, root mean square error, Kappa coefficient and AUROC value in a training set and a test set according to results, the model performance difference is obtained, the precision comparison of different algorithms is carried out, and the running performances of different algorithms are compared.
According to the technical scheme, in the fifth step, the mountain torrent easiness-to-send drawing means that a model result of the mountain torrent easiness-to-send degree of the small watershed is imported into ArcGIS10.4 software, and the mountain torrent disaster easiness is visually displayed.
According to the technical scheme, the mountain torrent easiness mapping comprises mapping and analyzing mountain torrent easiness results of the small watersheds, a predicted value is automatically output to each small watersheds through three machine learning models, and the output predicted value is the mountain torrent disaster easiness of the small watersheds;
after the vulnerability of the small watershed units is obtained, the result of the vulnerability is graded by adopting a natural breakpoint method and directly graded into five grades, and after grading, the number, the occupied area and the occupied proportion of the small watersheds of each grade are obtained through statistical analysis.
Compared with the prior art, the invention has the beneficial effects that:
1. the method applies a multivariate self-adaptive regression spline and a random forest algorithm as an integrated learning basic model, adopts an artificial neural network algorithm for integration, takes a small flood area vector unit as a research object, selects a mountain flood characteristic factor as a regression independent variable, and performs mountain flood easiness evaluation and mapping, can effectively combine mountain flood disaster investigation evaluation results with an advanced machine learning algorithm to perform mountain flood disaster easiness analysis, is different from other researches which take a grid as a basic unit, and can more accurately analyze the easiness degree of the small flood area by taking the vector small flood area as a basic research unit, thereby accurately positioning the small flood area where the mountain flood is easy to send, and can effectively combine the mountain flood disaster investigation evaluation results with machine learning and an intelligent algorithm by using the machine learning, thereby performing mountain flood disaster easiness evaluation and mapping.
Meanwhile, the evaluation method provided by the invention has the advantages of accurate calculation, reliable principle and easy and simple operation process, can be universally applied to the requirement of provincial scale mountain torrent disaster easiness evaluation, can be popularized to national mountain torrent disaster easiness evaluation in the future, and provides an auxiliary decision for prevention and treatment of mountain torrent disasters.
2. In the evaluation method, the small watershed for investigation and evaluation of the mountain torrent disasters is used as a basic research unit, so that the mountain torrent disaster easiness of the small watershed area can be evaluated more accurately, a new thought and direction are provided for future research, in the technical field, a machine learning model and an intelligent algorithm are used for researching the nonlinear influence of small watershed parameters on the mountain torrent disaster easiness, the research result provides a theoretical basis for further applying GIS to mountain torrent disaster evaluation and drawing, and in the practical application, the method can be effectively used as a visualization result for providing support and auxiliary decision for establishing a mountain torrent disaster early warning and prevention system for national defense and local governments.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention.
In the drawings:
FIG. 1 is a flow chart of the steps of the present invention;
FIG. 2 is a graphical illustration of single factor contribution in the RF model of the present invention;
FIG. 3 is a schematic representation of the impact of interaction factors within the MARS model of the present invention;
FIG. 4 is a schematic structural view of the present invention;
FIG. 5 is a schematic structural view of the present invention;
fig. 6 is a schematic structural view of the present invention.
Detailed Description
The preferred embodiments of the present invention will be described in conjunction with the accompanying drawings, and it will be understood that they are described herein for the purpose of illustration and explanation and not limitation.
Example (b): as shown in fig. 1, the invention provides a technical solution, which is a mountain torrent disaster proneness evaluation method based on GIS and ensemble learning, and applies a Multivariate Adaptive Regression Spline (MARS) and a Random Forest (RF) algorithm as an ensemble learning base model, and adopts Artificial Neural Network (ANN) algorithm integration, and 3860 small watershed vector units are used as research objects, and 13 mountain torrent characteristic factors are selected as regression independent variables;
the evaluation method specifically comprises the following steps:
firstly, preprocessing data;
secondly, establishing a model;
thirdly, the characteristic factors influence;
fourthly, verifying the performance of the algorithm;
and fifthly, drawing the mountain torrents easily.
Based on the technical scheme, in the first step, data preprocessing refers to preprocessing a small watershed data set, an evaluation index system is established according to the action mechanism of small watershed parameters on torrential flood disasters and according to an index selection principle, and indexes are considered from the three aspects of geometric characteristics (A1), hydrological characteristics (A2) and meteorological characteristics (A3).
Based on the technical scheme, the geometric characteristics (A1) are selected from an average slope (A11), a weighted average slope (A12), a shape coefficient (A13), a longest confluence path length (A14), a longest confluence path slope (A15) and a longest confluence path slope 1085 (A16);
selecting roughness (A21), stable infiltration rate (A22) and unit line peak modulus (A23) with the frequency of 1% in 10 minutes from hydrological characteristics (A2);
the meteorological characteristics (A3) are selected from unit line data (A31) with the frequency of 1% in 10 minutes, rainfall with the frequency of 1% in 10 minutes (A32), unit line data (A33) with the frequency of 1% in 30 minutes and rainfall with the frequency of 1% in 30 minutes (A24), and formed index systems and field abbreviations are shown in the following table:
index system for vulnerability of mountain torrent disasters
On the basis of data cleaning, standardization and downsampling, a small watershed data set is trained, and 70% of samples are randomly selected as a training set and 30% of samples are selected as a test set to be used as model input.
Based on the technical scheme, in the second step, the model establishment means that the selected 13 characteristic factors are used, two machine learning algorithms and an integrated algorithm are adopted to evaluate the easiness of the small watershed torrential flood disasters, and the method specifically comprises the following steps:
firstly, using a training set to initially train an algorithm;
then, in order to enable the algorithm to achieve the best performance, a Bayesian optimization algorithm and test set data are adopted to optimize the parameters;
finally, the indexes are evaluated by using the test set data calculation algorithm, and the lifting algorithms are compared in detail.
Based on the technical scheme, in order to overcome the unbalanced problem in mountain torrent proneness modeling and ensure that the proportion of '0' and '1' in a machine learning training model is the same, 1930 adding models are randomly selected in other small watersheds where mountain torrents do not occur, training sets and testing sets are respectively and randomly divided on mountain torrent and non-mountain torrent data sets according to the proportion of 7: 3, and 2772 training sets and 1088 testing sets of data are obtained.
Based on the technical scheme, in the second step, the model building comprises the steps of finishing the primary building of the MARS model by using the R language, finishing the primary building of the RF model by using the R language, and finishing the primary building of the ANN model by using the R language;
the primary establishment of the MARS model is completed by using the R language: in R4.0.4, a package of the MARS function is provided, the name is "earth", and after the package is loaded, a MARS model is built according to the requirements:
parameter optimization of a MARS model, wherein two tuning parameters are required to be set by the MARS method, the two tuning parameters are respectively called as degree and nprene for predicting the order of a variable and the number of terms for reserving characteristics, the degree is an integer which is not less than 1, the degree sets an upper limit which is not more than 3, the calculated amount is greatly increased due to a higher order, and the numerical value is greatly expanded or reduced, so that the value is not suitable to be overlarge, the nprene generally takes a value which is not less than 2, the tuning parameter degree is set to be 1-5 after comprehensive analysis, and the nprene is set to be 5-30;
the learning performance is measured by using the root mean square error, two parameters are selected when the RMSE is minimum, and the model obtains the best effect through experiments when nprun is 3 and degree is 14;
the primary building of the RF model is done in the R language: providing a program package of an RF function in R4.0.4, wherein the name is RandomForest, and after the program package is loaded, establishing an RF model according to requirements;
determining an establishing mode and selected parameters of an RF model, wherein the RF method needs to set two tuning parameters, ntree refers to the number of trees in a forest, mtry represents the number of characteristics used by each tree, the mtry node value can be generally defaulted to 2 or one third of the number of data set variables, the value of ntree can be determined after the mtry node value is determined, the preset range is 200-1000, and the parameters can be rapidly tuned by using For circulation;
and (3) completing the initial establishment of an ANN model by using an R language: with MARS and RF model predictions as inputs, a package of ANN functions, named "neuralenet", is provided at R4.0.4, and the ANN model is built after loading the package.
As shown in fig. 2 and 3, in the third step, the influence of the characteristic factor refers to the contribution degree of the characteristic factor to the easiness of the mountain torrent disaster;
obtaining the influence of single contribution of the characteristic factors and double-factor interaction on the mountain torrent disaster easiness according to results of the MARS and the RF model, mainly checking the importance of the factors in the RF model and checking the influence of multi-factor interaction in the MARS model on the mountain torrent easiness;
as can be seen in fig. 2: the first three factors of the single factor contribution degree rank are respectively: the unit line with the frequency of 1% in 10 minutes, the ratio drop of the longest confluence path and the rainfall with the frequency of 1% in 10 minutes can prove that short-duration heavy rainfall and terrain gradient are important factors influencing the easiness of mountain torrent disasters;
as can be seen in fig. 3: three pairs of double factors have interactive influence on the easiness of the mountain torrent disasters, wherein the most obvious two factors are an average slope and rainfall with the frequency of 1% in 10 minutes, and the short-duration heavy rainfall and the terrain slope are proved to be important factors influencing the easiness of the mountain torrent disasters.
Based on the technical scheme, in the fourth step, the verification algorithm performance refers to model verification and comparison, the precision of the algorithm is verified mainly by running test set data, in the performance testing part of the algorithm, the precision is evaluated by calculating algorithm evaluation indexes including precision, recall rate, precision, root mean square error, Kappa coefficient and AUROC value in a training set and a test set according to results, the performance difference of the models is obtained, the precision comparison of different algorithms is carried out, the running performances of different algorithms are compared, and the model performance evaluation is shown in the following table:
evaluation of model Performance
Based on the technical scheme, in the fifth step, the mountain torrent easiness in occurrence mapping means that a model result of the mountain torrent easiness in occurrence degree of the small watershed is imported into ArcGIS10.4 software, and the mountain torrent disaster easiness is visually displayed.
The method comprises the steps of mapping and analyzing the mountain torrent easiness result of the small watershed, automatically outputting a predicted value to each small watershed through three machine learning models, wherein the output predicted value is the mountain torrent disaster easiness of the small watershed, and the range of the predicted value is 0-1;
as shown in fig. 4, 5, and 6, after the vulnerability of the small watershed units is obtained, the result of vulnerability is graded by using a natural breakpoint method, and is directly graded into five grades, and after grading, the number, occupied area, and occupied ratio of the small watersheds of each grade are obtained through statistical analysis, such as shown in the following table:
distribution condition of vulnerability of small flow area, number/number, area/km2
Finally, it should be noted that: although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that changes may be made in the embodiments and/or equivalents thereof without departing from the spirit and scope of the invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A mountain torrent disaster easiness evaluation method based on GIS and ensemble learning is characterized in that: applying an ensemble learning algorithm, taking a plurality of small watershed vector units as research objects, and performing mountain torrent disaster proneness evaluation and drawing;
the method specifically comprises the following steps:
firstly, preprocessing data;
secondly, establishing a model;
thirdly, the characteristic factors influence;
fourthly, verifying the performance of the algorithm;
and fifthly, drawing the mountain torrents easily.
2. The mountain torrent disaster vulnerability assessment method based on GIS and ensemble learning of claim 1, wherein: in the first step, the data preprocessing refers to preprocessing a small watershed data set, an evaluation index system is established according to the action mechanism of small watershed parameters on torrential flood disasters and according to an index selection principle, and indexes are considered from the three aspects of geometric characteristics (A1), hydrological characteristics (A2) and meteorological characteristics (A3).
3. The mountain torrent disaster vulnerability assessment method based on GIS and ensemble learning of claim 2, wherein: selecting an average gradient (A11), a weighted average gradient (A12), a shape coefficient (A13), a longest confluence path length (A14), a longest confluence path slope (A15) and a longest confluence path slope 1085(A16) from the geometric characteristics (A1);
selecting roughness (A21), stable infiltration rate (A22) and unit line peak modulus (A23) with the frequency of 1% in 10 minutes from hydrological characteristics (A2);
selecting unit line data (A31) with the frequency of 1% in 10 minutes, rainfall (A32) with the frequency of 1% in 10 minutes, unit line data (A33) with the frequency of 1% in 30 minutes and rainfall (A24) with the frequency of 1% in 30 minutes from meteorological characteristics (A3);
on the basis of data cleaning, standardization and downsampling, a small watershed data set is trained, and 70% of samples are randomly selected as a training set and 30% of samples are selected as a test set to be used as model input.
4. The mountain torrent disaster vulnerability assessment method based on GIS and ensemble learning of claim 1, characterized in that: in the second step, the model building means that the selected characteristic factors are used, two machine learning algorithms and an integration algorithm are adopted to evaluate the easiness of the small watershed torrential flood disasters, and the method specifically comprises the following steps:
firstly, using a training set to initially train an algorithm;
then, in order to enable the algorithm to achieve the best performance, an optimization algorithm, a model initial result and test set data are adopted to optimize parameters;
finally, the indexes are evaluated by using the test set data calculation algorithm, and the lifting algorithms are compared in detail.
5. The mountain torrent disaster vulnerability assessment method based on GIS and ensemble learning of claim 4, wherein: in order to solve the unbalanced problem in mountain torrent proneness modeling and ensure the same proportion in a machine learning training model, after randomly selecting the same number of small watersheds from other small watersheds without mountain torrents and adding the small watersheds into the model, respectively randomly dividing a training set and a test set on a mountain torrent data set and a non-mountain torrent data set according to the proportion of 7: 3 to obtain a plurality of data training sets and a plurality of data test sets.
6. The mountain torrent disaster vulnerability assessment method based on GIS and ensemble learning of claim 4, wherein: in the second step, the model building comprises the steps of completing the primary building of a MARS model by using the R language, completing the primary building of an RF model by using the R language, and completing the primary building of an ANN model by using the R language.
7. The mountain torrent disaster vulnerability assessment method based on GIS and ensemble learning of claim 1, wherein: in the third step, the influence of the characteristic factors refers to the contribution degree of the characteristic factors to the easiness of the mountain torrent disasters;
the influence of single contribution of the characteristic factors and the influence of double-factor interaction on the mountain torrent disaster easiness are obtained according to results of the MARS model and the RF model, and the influence of factor importance in the RF model and the influence of multi-factor interaction in the MARS model on the mountain torrent disaster easiness are mainly checked.
8. The mountain torrent disaster vulnerability assessment method based on GIS and ensemble learning of claim 1, wherein: in the fourth step, the verification algorithm performance refers to model verification and comparison, the accuracy of the algorithm is verified mainly by running test set data, and in the test algorithm performance part, the accuracy including the accuracy, the recall rate, the accuracy, the root mean square error, the Kappa coefficient and the AUROC value is evaluated by calculating algorithm evaluation indexes in the training set and the test set according to results, so that model performance difference is obtained, the accuracy comparison of different algorithms is carried out, and the running performances of different algorithms are compared.
9. The mountain torrent disaster vulnerability assessment method based on GIS and ensemble learning of claim 1, wherein: in the fifth step, the mountain torrent easiness in occurrence mapping means that a model result of the mountain torrent easiness in occurrence degree of the small watershed is imported into ArcGIS10.4 software, and the mountain torrent disaster easiness in occurrence is visually displayed.
10. The mountain torrent disaster vulnerability assessment method based on GIS and ensemble learning of claim 9, wherein: the mountain torrent easiness mapping comprises mapping and analyzing mountain torrent easiness results of the small watershed, a predicted value is automatically output to each small watershed through three machine learning models, and the output predicted value is the mountain torrent disaster easiness of the small watershed;
after the vulnerability of the small watershed units is obtained, the result of the vulnerability is graded by adopting a natural discontinuity method and directly graded into five grades, and after grading, the number, the occupied area and the occupied proportion of the small watersheds of each grade are obtained through statistical analysis.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210078561.8A CN114493245A (en) | 2022-01-21 | 2022-01-21 | Mountain torrent disaster easiness evaluation method based on GIS and integrated learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210078561.8A CN114493245A (en) | 2022-01-21 | 2022-01-21 | Mountain torrent disaster easiness evaluation method based on GIS and integrated learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114493245A true CN114493245A (en) | 2022-05-13 |
Family
ID=81474315
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210078561.8A Pending CN114493245A (en) | 2022-01-21 | 2022-01-21 | Mountain torrent disaster easiness evaluation method based on GIS and integrated learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114493245A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115660424A (en) * | 2022-10-28 | 2023-01-31 | 国网四川省电力公司 | Disaster factor analysis early warning system based on GIS |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108280553A (en) * | 2018-02-24 | 2018-07-13 | 中山大学 | Regional Torrent Risk Zonation based on GIS- Artificial neural network ensembles and prediction technique |
| CN111968019A (en) * | 2020-07-31 | 2020-11-20 | 中国地质大学（武汉） | Landslide hazard susceptibility prediction method based on CNN and RNN integration |
| CN113191582A (en) * | 2021-03-15 | 2021-07-30 | 西南石油大学 | Road torrential flood vulnerability evaluation method based on GIS and machine learning |
-
2022
- 2022-01-21 CN CN202210078561.8A patent/CN114493245A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108280553A (en) * | 2018-02-24 | 2018-07-13 | 中山大学 | Regional Torrent Risk Zonation based on GIS- Artificial neural network ensembles and prediction technique |
| CN111968019A (en) * | 2020-07-31 | 2020-11-20 | 中国地质大学（武汉） | Landslide hazard susceptibility prediction method based on CNN and RNN integration |
| CN113191582A (en) * | 2021-03-15 | 2021-07-30 | 西南石油大学 | Road torrential flood vulnerability evaluation method based on GIS and machine learning |
Non-Patent Citations (3)
| Title |
|---|
| 吴小君;方秀琴;任立良;吴陶樱;苗月鲜;: "基于随机森林的山洪灾害风险评估――以江西省为例", 水土保持研究, no. 03, 24 April 2018 (2018-04-24), pages 142 - 149 * |
| 管珉;陈兴旺;: "江西省山洪灾害风险区划初步研究", 暴雨灾害, no. 04, 15 December 2007 (2007-12-15), pages 339 - 342 * |
| 葛星;骆建雄;: "GIS支持下的小流域山洪灾害风险区划分方法研究", 中国农村水利水电, no. 10, 15 October 2018 (2018-10-15), pages 170 - 176 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115660424A (en) * | 2022-10-28 | 2023-01-31 | 国网四川省电力公司 | Disaster factor analysis early warning system based on GIS |
| CN115660424B (en) * | 2022-10-28 | 2024-02-13 | 国网四川省电力公司 | Disaster element analysis early warning system based on GIS |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108280553B (en) | Mountain torrent disaster risk zoning and prediction method based on GIS-neural network integration | |
| Olauson | ERA5: The new champion of wind power modelling? | |
| CN106651211B (en) | Flood disaster risk assessment method for different scale areas | |
| US20170261646A1 (en) | Self-correcting multi-model numerical rainfall ensemble forecasting method | |
| CN111665575B (en) | Medium-and-long-term rainfall grading coupling forecasting method and system based on statistical power | |
| CN109242149A (en) | A kind of student performance early warning method and system excavated based on educational data | |
| CN107633320B (en) | Power grid line importance degree evaluation method based on meteorological prediction and risk evaluation | |
| CN116451879A (en) | Drought risk prediction method and system and electronic equipment | |
| CN112285807A (en) | Meteorological information prediction method and device | |
| CN111178585A (en) | Fault reporting amount prediction method based on multi-algorithm model fusion | |
| CN115270527B (en) | Real-time assessment method, equipment and storage medium for road collapse risk | |
| CN116609860B (en) | Hydrologic model real-time correction method and system based on integrated learning algorithm | |
| CN114493052B (en) | Multi-model fusion self-adaptive new energy power prediction method and system | |
| CN111968019A (en) | Landslide hazard susceptibility prediction method based on CNN and RNN integration | |
| Hou et al. | Damage prediction of 10 kV power towers in distribution network under typhoon disaster based on data-driven model | |
| Yang et al. | Resilience assessment and improvement for electric power transmission systems against typhoon disasters: a data-model hybrid driven approach | |
| CN116777079A (en) | Desertification intrusion disaster prediction method based on Bayesian interlayer structure model | |
| CN113642699A (en) | Intelligent river flood forecasting system | |
| CN115544889A (en) | Numerical mode precipitation deviation correction method based on deep learning | |
| AU2019100631A4 (en) | Self-correcting multi-model numerical rainfall ensemble forecasting method | |
| CN114493245A (en) | Mountain torrent disaster easiness evaluation method based on GIS and integrated learning | |
| CN111445087A (en) | Flood prediction method based on extreme learning machine | |
| CN116976227B (en) | Storm water increasing forecasting method and system based on LSTM machine learning | |
| CN117454059A (en) | Runoff calculation and prediction method based on watershed hydrologic model | |
| CN112989557A (en) | Method for improving water reserve change prediction reliability based on neural network selectable model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |