CN117556714B - Preheating pipeline temperature data anomaly analysis method for aluminum metal smelting - Google Patents
Preheating pipeline temperature data anomaly analysis method for aluminum metal smelting Download PDFInfo
- Publication number
- CN117556714B CN117556714B CN202410043822.1A CN202410043822A CN117556714B CN 117556714 B CN117556714 B CN 117556714B CN 202410043822 A CN202410043822 A CN 202410043822A CN 117556714 B CN117556714 B CN 117556714B
- Authority
- CN
- China
- Prior art keywords
- abnormal
- value
- local
- obtaining
- local extremum
- 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.)
- Active
Links
- 238000003723 Smelting Methods 0.000 title claims abstract description 22
- XAGFODPZIPBFFR-UHFFFAOYSA-N aluminium Chemical compound [Al] XAGFODPZIPBFFR-UHFFFAOYSA-N 0.000 title claims abstract description 22
- 238000004458 analytical method Methods 0.000 title claims abstract description 21
- 230000002159 abnormal effect Effects 0.000 claims abstract description 138
- 238000012544 monitoring process Methods 0.000 claims abstract description 83
- 238000001514 detection method Methods 0.000 claims abstract description 43
- 238000005457 optimization Methods 0.000 claims abstract description 15
- 238000012512 characterization method Methods 0.000 claims description 36
- 238000009826 distribution Methods 0.000 claims description 36
- 238000000034 method Methods 0.000 claims description 23
- 230000005856 abnormality Effects 0.000 claims description 12
- 238000013507 mapping Methods 0.000 claims description 3
- 238000012216 screening Methods 0.000 description 5
- 230000002902 bimodal effect Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01K—MEASURING TEMPERATURE; MEASURING QUANTITY OF HEAT; THERMALLY-SENSITIVE ELEMENTS NOT OTHERWISE PROVIDED FOR
- G01K13/00—Thermometers specially adapted for specific purposes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/08—Thermal analysis or thermal optimisation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention relates to the technical field of data anomaly detection, in particular to a preheating pipeline temperature data anomaly analysis method for aluminum metal smelting; obtaining a local extremum curve of the abnormal fraction of the temperature monitoring data set according to the isolated forest model; obtaining a fluctuation factor, a difference factor and a change factor according to the data characteristics in the local extremum curve; obtaining the recommended degree, the recommended threshold and the detection accuracy of the local minimum point according to the fluctuation factor, the difference factor and the change factor; and obtaining an optimal recommended threshold according to the recommended threshold and the detection accuracy of the isolated forest model under different temperature monitoring data sets. According to the invention, the isolated forest model is iterated based on the optimal recommended threshold value to obtain the isolated forest optimization model, and the isolated forest optimization model is utilized to perform the temperature anomaly analysis of the preheating pipeline, so that the anomaly analysis accuracy is improved.
Description
Technical Field
The invention relates to the technical field of data anomaly detection, in particular to a preheating pipeline temperature data anomaly analysis method for aluminum metal smelting.
Background
In the aluminum metal smelting process, the temperature monitoring of the preheating pipeline is important, and the operation state of the preheating pipeline directly influences the production efficiency and even damages equipment. In order to ensure the accuracy of the result when the temperature is abnormally detected, errors caused by only analyzing temperature data are avoided, so that monitoring is required to be carried out from operation characteristics of multiple dimensions, such as pressure and flow of a pipe affected by the temperature, and the temperature abnormality is judged according to the data analysis of the multiple dimensions.
The existing isolated forest algorithm is generally used for carrying out abnormality monitoring on multi-dimensional data, a plurality of isolated trees are constructed in an isolated forest, an abnormality score is obtained by calculating the average path length of all the isolated trees, and an abnormality sample is screened through a threshold value of the abnormality score so as to realize temperature abnormality detection. The value selection of the threshold value influences the screening result of the abnormal sample, and the monitoring data have different degrees of fluctuation due to different equipment operation conditions in different time periods, so that the abnormal data are difficult to accurately distinguish through the empirically selected threshold value, and the accuracy of abnormal detection of the temperature data is influenced.
Disclosure of Invention
In order to solve the technical problem that the abnormal data are difficult to accurately distinguish by the empirically selected threshold value and the accuracy of temperature data abnormal detection is affected, the invention aims to provide a preheating pipeline temperature data abnormal analysis method for aluminum metal smelting, and the adopted technical scheme is as follows:
acquiring a temperature monitoring data set of a preheating pipeline in the aluminum metal smelting process; acquiring an abnormal distribution curve of the abnormal fraction of the temperature monitoring data set according to the isolated forest model; obtaining a local extremum curve according to the data change characteristics of the abnormal distribution curve;
obtaining a fluctuation factor according to the data fluctuation characteristics in the preset neighborhood of the local minimum point in the local extremum curve; obtaining a difference factor according to the difference characteristics of the local minimum point and other local extremum points in the local extremum curve; obtaining a change factor according to the discrete characteristics of the local minimum point and the adjacent local extremum point in the local extremum curve;
obtaining the recommended degree of the local minimum point according to the fluctuation factor, the difference factor and the change factor; acquiring a recommendation threshold value of the isolated forest model according to recommendation degrees of all local minimum points; obtaining the detection accuracy of the isolated forest model according to the recommended threshold;
acquiring recommended thresholds and detection accuracy of the isolated forest model under different temperature monitoring data sets; acquiring an optimal recommendation threshold of the isolated forest model according to all recommendation thresholds and detection accuracy; and carrying out iteration on the isolated forest model based on the optimal recommended threshold value to obtain an isolated forest optimization model, and carrying out temperature anomaly analysis on the preheating pipeline by using the isolated forest optimization model.
Further, the step of obtaining an anomaly profile of the anomaly score of the temperature monitoring dataset from the isolated forest model includes:
obtaining abnormal fractions of each temperature monitoring sample in the temperature monitoring data set according to the isolated forest model; and constructing a two-dimensional rectangular coordinate system of an abnormal distribution curve according to the abnormal scores of the temperature monitoring samples, wherein the transverse axis of the two-dimensional rectangular coordinate system of the abnormal distribution curve is the abnormal scores sequenced from small to large, the vertical axis is the number value of the temperature monitoring samples corresponding to the abnormal scores, and connecting the number values of the temperature monitoring samples corresponding to each abnormal distribution to obtain the abnormal distribution curve.
Further, the step of obtaining a local extremum curve according to the data change characteristic of the abnormal distribution curve comprises the following steps:
and obtaining local extremum points in the abnormal distribution curve, constructing a two-dimensional rectangular coordinate system related to the local extremum curve, wherein the horizontal axis of the two-dimensional rectangular coordinate system of the local extremum curve is an abnormal fraction corresponding to the local extremum points, the vertical axis is a numerical value of the local extremum points, and connecting the numerical values of each local extremum point to obtain the local extremum curve.
Further, the step of obtaining the fluctuation factor according to the data fluctuation characteristics in the preset neighborhood of the local minimum point in the local extremum curve includes:
calculating the absolute value of the slope of a straight line between a local extremum point and an adjacent local extremum point in the local extremum curve, and obtaining the slope representation value of the local extremum point; and calculating the variance of the slope representation value of the local extreme point in the preset neighborhood of the local minimum point to obtain the fluctuation factor of the local minimum point.
Further, the step of obtaining the difference factor according to the difference characteristics of the local minimum point and other local extremum points in the local extremum curve includes:
calculating the numerical average value of local extremum points in the local extremum curve to obtain an integral characterization value; calculating the numerical value difference between the integral characterization value and the local minimum value point to obtain an integral difference value; when the integral difference value is a positive number, calculating the ratio of the integral difference value to the integral characterization value to obtain a difference factor of the local minimum point; and when the overall difference value is not positive, the difference factor of the local minimum value point is a preset first constant.
Further, the step of obtaining the change factor according to the discrete features of the local minimum point and the adjacent local extremum point in the local extremum curve includes:
and calculating the average value of the slope characterization values of the local minimum points and the adjacent local extremum points to obtain the change factor of the local minimum.
Further, the step of obtaining the recommended degree of the local minimum point according to the fluctuation factor, the difference factor and the change factor includes:
and calculating the product of the fluctuation factor, the difference factor and the change factor of the local minimum point to obtain the recommended degree of the local minimum point.
Further, the step of obtaining the recommendation threshold of the isolated forest model according to the recommendation degree of all the local minimum points comprises the following steps:
and taking the abnormal score corresponding to the local minimum point of the recommended degree maximum value in the local extremum curve as a recommended threshold value of the isolated forest model.
Further, the step of obtaining the detection accuracy of the isolated forest model according to the recommendation threshold includes:
obtaining abnormal samples in the temperature monitoring data set through an isolated forest model according to the recommended threshold value, and counting the number of actual abnormal samples in the abnormal samples to obtain abnormal quantity; and calculating the ratio of the abnormal quantity to the number of the abnormal samples to obtain the detection accuracy of the isolated forest model.
Further, the step of obtaining the optimal recommendation threshold of the isolated forest model according to all recommendation thresholds and detection accuracy comprises the following steps:
calculating the average value of the absolute values of the differences between the recommendation threshold and other recommendation thresholds, and obtaining a central characterization value of the recommendation threshold; taking the recommended threshold with the maximum central characterization value as a reference recommended threshold; calculating the absolute value of the difference between any recommended threshold and a reference recommended threshold and carrying out negative correlation mapping to obtain a distance characterization value of the any recommended threshold; and calculating multiplication results among the arbitrary recommendation threshold, the distance characterization value and the detection accuracy aiming at the arbitrary recommendation threshold, and taking the average value of the multiplication results corresponding to all recommendation thresholds as the optimal recommendation threshold of the isolated forest model.
The invention has the following beneficial effects:
in the invention, the acquired local extremum curve can reflect the number of the temperature monitoring samples corresponding to different abnormal scores, so that the recommended threshold value can be conveniently selected; the fluctuation factors of the local minimum points can be obtained to select proper recommendation thresholds based on the fluctuation characteristics of the data, the difference factors can be obtained to select proper recommendation thresholds according to the difference characteristics of the data, and the change factors can be obtained to select proper recommendation thresholds according to the discrete characteristics of the data; the recommendation degree can be obtained by accurately selecting the recommendation threshold value on the basis of combining the fluctuation factor, the difference factor and the change factor. Obtaining the detection accuracy can reflect the suitability of the recommended threshold; the purpose of obtaining the recommended threshold values and the detection accuracy corresponding to different temperature monitoring data sets is to increase the reliability of the recommended threshold values by combining the running processes of different preheating pipelines, so as to obtain the optimal recommended threshold values. And finally, an isolated forest optimization model is obtained based on the optimal recommended threshold value, and the abnormal analysis of the temperature is carried out, so that the accuracy of the abnormal analysis of the temperature is improved.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method for analyzing the temperature data anomalies of a preheating pipeline for aluminum metal smelting according to an embodiment of the present invention.
Detailed Description
In order to further describe the technical means and effects adopted by the invention to achieve the preset aim, the following description refers to the specific implementation, structure, characteristics and effects of a preheating pipeline temperature data anomaly analysis method for aluminum metal smelting according to the invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The invention provides a specific scheme of a preheating pipeline temperature data anomaly analysis method for aluminum metal smelting, which is specifically described below with reference to the accompanying drawings.
Referring to fig. 1, a flowchart of a preheating pipeline temperature data anomaly analysis method for aluminum metal smelting according to an embodiment of the present invention is shown, the method includes the following steps:
step S1, acquiring a temperature monitoring data set of a preheating pipeline in the aluminum metal smelting process; acquiring an abnormal distribution curve of the abnormal fraction of the temperature monitoring data set according to the isolated forest model; and obtaining a local extremum curve according to the data change characteristic of the abnormal distribution curve.
In the embodiment of the invention, the implementation scene is to detect the abnormality of the preheating pipeline temperature in aluminum metal smelting; firstly, acquiring a temperature monitoring data set of a preheating pipeline in the aluminum metal smelting process, and if only analyzing the temperature data, possibly causing data abnormality due to noise or sensor faults, affecting the reliability of abnormality detection; therefore, the characteristics which are easy to be affected by temperature, such as the pressure and the flow in the preheating pipeline, the numerical values of the pressure and the flow in the preheating pipeline at different temperatures are different, the pressure data, the flow data and the temperature data of different positions of the preheating pipeline are obtained, a temperature monitoring data set of multi-dimensional data is constructed, the multi-dimensional characteristics at one acquisition time are used as a temperature monitoring sample, and the temperature monitoring data set comprises a normal sample and an abnormal sample; the practitioner can determine the acquisition object and the acquisition frequency according to the implementation scene.
The isolated forest algorithm is an existing anomaly detection algorithm, an anomaly score of each sample is obtained by constructing a plurality of isolated trees and calculating the average path length of the sample, and finally an anomaly sample is screened out by empirically setting a threshold value of the anomaly score. The detection accuracy of the abnormal samples is related to the selection of the threshold value, the abnormal samples are omitted when the threshold value is selected too high, otherwise, the abnormal samples are mistakenly considered as the normal data when the threshold value is selected too low; therefore, in order to improve the accuracy of the isolated forest model for detecting the abnormity of the preheating pipeline, the selection of the threshold value needs to be improved.
Firstly, acquiring an abnormal distribution curve of abnormal scores of a temperature monitoring dataset according to an isolated forest model, wherein the abnormal distribution curve specifically comprises the following steps: obtaining abnormal fractions of each temperature monitoring sample in the temperature monitoring data set according to the isolated forest model; and constructing a two-dimensional rectangular coordinate system of the abnormal distribution curve according to the abnormal scores of the temperature monitoring samples, wherein the transverse axis of the two-dimensional rectangular coordinate system of the abnormal distribution curve is the abnormal score sequenced from small to large, the vertical axis is the number value of the temperature monitoring samples corresponding to the abnormal score, and connecting the number values of the temperature monitoring samples corresponding to each abnormal distribution to obtain the abnormal distribution curve. It should be noted that, the abnormal score obtained by obtaining the temperature monitoring sample through the isolated forest model belongs to the prior art, and specific calculation steps are not repeated. If a part of obvious abnormal temperature monitoring samples exist in the temperature monitoring data set, obvious double-peak distribution characteristics are displayed in an abnormal distribution curve, wherein a larger peak is the number of normal temperature monitoring samples represented by a lower abnormal fraction, and a smaller peak is the number of abnormal temperature monitoring samples represented by a larger abnormal fraction; for temperature monitoring datasets where significant anomalies exist, the trough between the peaks in the anomaly profile can be used as a threshold. However, for a temperature monitoring data set with unobvious abnormal samples, the corresponding bimodal distribution characteristics of the abnormal distribution curve are weak, and it is difficult to select an accurate threshold value to screen the abnormal samples. Therefore, the threshold selection needs to be analyzed according to the change characteristics of the abnormal distribution curve, and a local extremum curve is obtained according to the data change characteristics of the abnormal distribution curve.
Further, in the embodiment of the present invention, the step of obtaining the local extremum curve includes: obtaining local extremum points in an abnormal distribution curve, wherein the local extremum points comprise local maximum points and local minimum points; when any data point is smaller than two data points adjacent to the left and right, the any data point is a local minimum value point, and when any data point is larger than two data points adjacent to the left and right, the any data point is a local maximum value point. Constructing a two-dimensional rectangular coordinate system about a local extremum curve, wherein the transverse axis of the two-dimensional rectangular coordinate system of the local extremum curve is an abnormal fraction corresponding to a local extremum point, and the distances between two adjacent local extremum points in the transverse axis are equal; and the vertical axis is the numerical value of the local extremum point, and the numerical values of the local extremum points are connected to obtain a local extremum curve. The local extremum curve reflects the abnormal fraction of the temperature monitoring sample with more and less existence, and the analysis can be carried out according to the characteristics of the local extremum curve, so that a proper threshold value is selected.
S2, obtaining a fluctuation factor according to the data fluctuation characteristics in the preset neighborhood of the local minimum point in the local extremum curve; obtaining a difference factor according to the difference characteristics of the local minimum points and other local extremum points in the local extremum curve; and obtaining a change factor according to the discrete characteristics of the local minimum point and the adjacent local extreme point in the local extreme curve.
In a temperature monitoring data set, a normal temperature monitoring sample accounts for most, an abnormal temperature monitoring sample accounts for less, a weaker bimodal distribution characteristic is shown in an abnormal distribution curve, the number of samples corresponding to the abnormal fraction of a transition part between the normal temperature monitoring sample and the abnormal temperature monitoring sample is smaller, and the abnormal distribution curve can be a local minimum point; therefore, the local minimum value point in the local extremum curve can be analyzed, and the degree to which the abnormal score corresponding to the local minimum value point becomes the threshold value is judged; firstly, obtaining a fluctuation factor according to the data fluctuation characteristics in the preset neighborhood of the local minimum point in the local extremum curve.
Preferably, in one embodiment of the present invention, the step of obtaining the fluctuation factor includes: calculating the absolute value of the slope of a straight line between a local extremum point and an adjacent local extremum point in the local extremum curve, and obtaining the slope representation value of the local extremum point; in the embodiment of the invention, a slope characterization value is calculated by using a local extremum point and a local extremum point adjacent to the last local extremum point; the slope representation value reflects the number difference characteristics of the temperature monitoring samples corresponding to the adjacent local extremum points, and when the slope representation value is larger, the number difference of the temperature monitoring samples corresponding to the two adjacent local extremum points is larger. Calculating the variance of the slope characterization value of the local extreme point in the preset neighborhood of the local minimum point to obtain the fluctuation factor of the local minimum point; in the abnormal distribution curve, the abnormal scores corresponding to the normal temperature monitoring samples are close, and the quantity value of the temperature monitoring samples corresponding to each abnormal score is close; the abnormal scores corresponding to the abnormal temperature monitoring samples are close, and the quantity value of the temperature monitoring samples corresponding to each abnormal score is close; if the local minimum point is in an abnormal score interval corresponding to an abnormal or normal temperature monitoring sample, the corresponding gradient characterization value difference is smaller, and the obtained fluctuation factor value is smaller; on the contrary, when the local minimum point is positioned in the middle of the abnormal scores corresponding to the abnormal and normal temperature monitoring samples, the difference of the quantity values of the temperature monitoring samples corresponding to the abnormal scores at the two ends of the local minimum point is larger, the difference of the slope characterization values is also larger, and the finally obtained fluctuation factor value is larger; therefore, when the fluctuation factor of the local minimum point is larger, the abnormal score corresponding to the local minimum point is more likely to be in the middle of the abnormal scores corresponding to the normal and abnormal temperature monitoring samples, and the abnormal score corresponding to the local minimum point is more suitable for being used as a screening threshold value of the abnormal samples. In the embodiment of the invention, the preset neighborhood is a range of 8 local extremum points which are left and right adjacent by taking the local extremum point as a center, and if the range rule is not satisfied, the range of the local extremum point closest to the center is taken as the preset neighborhood, so that an implementer can determine according to implementation scenes.
Further, in order to improve the accuracy of the threshold selection, the difference factor may be obtained continuously according to the difference characteristics of the local minimum point and other local extremum points in the local extremum curve, which specifically includes: calculating the numerical average value of local extremum points in the local extremum curve to obtain an integral characterization value; the overall characterization value reflects the average value of the number of temperature monitoring samples corresponding to each local extreme point. Calculating the numerical value difference between the integral characterization value and the local minimum value point to obtain an integral difference value; when the overall difference value is a positive number, the value of the temperature monitoring sample corresponding to the local minimum value point is smaller, and the value is possibly in the middle of the abnormal scores corresponding to the normal and abnormal temperature monitoring samples, and the ratio of the overall difference value to the overall characterization value is calculated to obtain the difference factor of the local minimum value point; when the ratio is larger, the difference factor is larger, which means that the smaller the number of temperature monitoring samples corresponding to the local minimum point is, the greater the possibility of being in the middle of abnormal scores corresponding to normal and abnormal temperature monitoring samples is, and the abnormal score corresponding to the local minimum point is suitable as a screening threshold value of the abnormal samples. When the overall difference value is not positive, it means that the temperature monitoring sample number corresponding to the local minimum point is higher than the average level, the horizontal axis of the local minimum point is on the abnormal score corresponding to the normal or abnormal temperature monitoring sample, and is not suitable for being used as the screening threshold of the abnormal sample, the difference factor of the local minimum point is set as a preset first constant, in the embodiment of the invention, the preset first constant is 0, and the implementer can determine according to implementation scenarios by himself.
According to the process of calculating the fluctuation factor, when the anomaly score corresponding to the local minimum point is more suitable for being used as a threshold value, the numerical variation of the local extreme points at the two ends of the local minimum point is more obvious, namely the variation of the number of temperature monitoring samples of the anomaly scores corresponding to the two ends is more obvious. Therefore, in order to further improve the accuracy of threshold selection, a change factor can be obtained according to the discrete characteristics of the local minimum point and the adjacent local extremum point in the local extremum curve; preferably, in one embodiment of the present invention, the step of obtaining the variation factor includes: calculating the average value of slope characterization values of a local minimum value point and an adjacent local extreme value point to obtain a change factor of the local minimum value, wherein the adjacent local extreme value point is the next local extreme value point of the local minimum value point; when the change factor is larger, the slope characterization value of the local minimum point and the adjacent local extremum point is larger, the difference between the number of temperature monitoring samples corresponding to the local minimum point and the local extremum point adjacent to the two ends is larger, and the abnormal score corresponding to the local minimum point is more likely to be used as a threshold value.
Step S3, obtaining the recommended degree of the local minimum point according to the fluctuation factor, the difference factor and the change factor; acquiring a recommendation threshold value of the isolated forest model according to recommendation degrees of all local minimum points; and obtaining the detection accuracy of the isolated forest model according to the recommended threshold.
The fluctuation factor, the difference factor and the change factor can reflect the possibility that the abnormal score corresponding to the local minimum point is used as the threshold value, so that the recommended degree of the local minimum point can be obtained according to the fluctuation factor, the difference factor and the change factor, and the method specifically comprises the following steps: calculating the product of the fluctuation factor, the difference factor and the change factor of the local minimum value point to obtain the recommended degree of the local minimum value point; when the fluctuation factor, the difference factor and the change factor of the local minimum point are all larger, the recommendation degree is larger, and the abnormal score of the local minimum point is more suitable to be used as a threshold value; the formula for obtaining the recommended degree comprises the following steps:
in the method, in the process of the invention,indicate->Recommended degree of individual local minima points, +.>Representing the number of local extreme points within the preset neighborhood of local extreme points, +.>Indicate->The first ∈of the preset neighborhood of the local minimum point>Slope representation value of individual local extremum points, +.>Indicate->The average value of the slope characterization values of the preset neighborhood of the local minimum points,indicate->A fluctuation factor of each local minimum point; />Representing the overall characterization value>Indicate->The numerical value of the local minimum point is described as +.>When the difference is positive, the difference factor isWhen the difference factor is not positive, the difference factor is a preset first constant; />Indicate->Slope-characterizing value of individual local minimum points, < ->Indicate->Slope characterization value of the local minimum point, i.e.>Slope representation value of adjacent local extremum point of local extremum points>Representing the variation factor.
The method for acquiring the recommendation degree of each local minimum point in the local extremum curve comprises the following steps of: and taking the abnormal score corresponding to the local minimum point of the maximum recommended degree in the local extremum curve as a recommended threshold value of the isolated forest model, and reasonably and accurately distinguishing the abnormal sample from the normal sample according to the recommended threshold value.
Further, after obtaining the recommended threshold, it is necessary to analyze the accuracy of anomaly detection passing through the recommended threshold, so that the accuracy of detection of the isolated forest model is obtained according to the recommended threshold, which specifically includes: according to the recommended threshold, obtaining an abnormal sample in the temperature monitoring data set through an isolated forest model, wherein the abnormal sample obtained through the isolated forest model belongs to the prior art, and the temperature monitoring sample with the abnormal score exceeding the recommended threshold is used as the abnormal sample by calculating the abnormal score of each temperature monitoring sample, so that the specific calculation steps are not repeated; counting the number of actual abnormal samples in the abnormal samples to obtain abnormal quantity, wherein the actual abnormal data are determined by an implementer; calculating the ratio of the abnormal quantity to the quantity of the abnormal samples to obtain the detection accuracy of the isolated forest model; the higher the detection accuracy means the more appropriate the recommendation threshold.
S4, acquiring recommended thresholds and detection accuracy of the isolated forest model under different temperature monitoring data sets; acquiring an optimal recommendation threshold of the isolated forest model according to all recommendation thresholds and detection accuracy; and carrying out iteration on the isolated forest model based on the optimal recommended threshold value to obtain an isolated forest optimization model, and carrying out temperature anomaly analysis on the preheating pipeline by using the isolated forest optimization model.
Because the currently acquired recommended threshold is calculated and analyzed according to one temperature monitoring data set, in order to improve the reliability of the recommended threshold, the temperature monitoring data sets of a plurality of different operation processes of the preheating pipeline are required to be prepared, and the recommended threshold and the detection accuracy of the isolated forest model under the different temperature monitoring data sets are acquired; obtaining an optimal recommendation threshold of the isolated forest model according to all recommendation thresholds and detection accuracy, wherein the optimal recommendation threshold comprises the following specific steps: calculating the average value of the absolute values of the differences between the recommendation threshold and other recommendation thresholds, and obtaining a central characterization value of the recommendation threshold; taking the recommended threshold with the smallest center characterization value as a reference recommended threshold; the minimum difference between the baseline recommended threshold and the other recommended thresholds means that the baseline recommended threshold is suitable as a threshold for anomaly detection in most temperature monitoring datasets. Calculating the absolute value of the difference between the random recommendation threshold and the reference recommendation threshold and carrying out negative correlation mapping to obtain a distance characterization value of the random recommendation threshold; the smaller the distance characterization value, the larger the difference between the arbitrary recommendation threshold and the reference recommendation threshold, the less suitable the arbitrary recommendation threshold is as a final threshold, and the smaller the weight of the arbitrary recommendation threshold is; conversely, the larger the distance characterization value, the more suitable the arbitrary recommendation threshold is as the final threshold, and the greater the weight of the arbitrary recommendation threshold. And aiming at any recommendation threshold, calculating multiplication results among the any recommendation threshold, the distance characterization value and the detection accuracy, taking the mean value of the multiplication results corresponding to all recommendation thresholds as the optimal recommendation threshold of the isolated forest model, and improving the accuracy of anomaly detection through the optimal recommendation threshold. The formula for obtaining the optimal recommendation threshold comprises the following steps:
in the method, in the process of the invention,representing an optimal recommended threshold,/->Indicates the number of recommended thresholds, +.>Indicate->The value of the recommended threshold,/>Indicate->Distance characterization value of the recommended threshold, +.>Indicate->The detection accuracy of each recommended threshold.
Further, the isolated forest model can be iterated based on the optimal recommendation threshold to obtain an isolated forest optimization model, in the embodiment of the invention, the isolated forest model is iterated according to the classification result of the optimal recommendation threshold, parameters in the model are optimized until the detection accuracy reaches 95%, the isolated forest optimization model is obtained, and an implementer can determine according to implementation scenes. Finally, carrying out temperature anomaly analysis on the preheating pipeline by using an isolated forest optimization model, calculating a temperature monitoring data set in the subsequent operation process of the preheating pipeline according to the isolated forest optimization model, and screening an anomaly sample according to an optimal recommended threshold; the abnormal sample can be detected more accurately according to the implementation scene of the isolated forest algorithm based on the preheating pipeline, and the accuracy of temperature data abnormal detection is improved.
In summary, the embodiment of the invention provides a preheating pipeline temperature data anomaly analysis method for aluminum metal smelting; obtaining a local extremum curve of the abnormal fraction of the temperature monitoring data set according to the isolated forest model; obtaining a fluctuation factor, a difference factor and a change factor according to the data characteristics in the local extremum curve; obtaining the recommended degree, the recommended threshold and the detection accuracy of the local minimum point according to the fluctuation factor, the difference factor and the change factor; and obtaining an optimal recommended threshold according to the recommended threshold and the detection accuracy of the isolated forest model under different temperature monitoring data sets. According to the invention, the isolated forest model is iterated based on the optimal recommended threshold value to obtain the isolated forest optimization model, and the isolated forest optimization model is utilized to perform the temperature anomaly analysis of the preheating pipeline, so that the anomaly analysis accuracy is improved.
It should be noted that: the sequence of the embodiments of the present invention is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments.
Claims (6)
1. The preheating pipeline temperature data anomaly analysis method for aluminum metal smelting is characterized by comprising the following steps of:
acquiring a temperature monitoring data set of a preheating pipeline in the aluminum metal smelting process; acquiring an abnormal distribution curve of the abnormal fraction of the temperature monitoring data set according to the isolated forest model; obtaining a local extremum curve according to the data change characteristics of the abnormal distribution curve;
obtaining a fluctuation factor according to the data fluctuation characteristics in the preset neighborhood of the local minimum point in the local extremum curve; obtaining a difference factor according to the difference characteristics of the local minimum point and other local extremum points in the local extremum curve; obtaining a change factor according to the discrete characteristics of the local minimum point and the adjacent local extremum point in the local extremum curve;
obtaining the recommended degree of the local minimum point according to the fluctuation factor, the difference factor and the change factor; acquiring a recommendation threshold value of the isolated forest model according to recommendation degrees of all local minimum points; obtaining the detection accuracy of the isolated forest model according to the recommended threshold;
acquiring recommended thresholds and detection accuracy of the isolated forest model under different temperature monitoring data sets; acquiring an optimal recommendation threshold of the isolated forest model according to all recommendation thresholds and detection accuracy; performing iteration on the isolated forest model based on the optimal recommended threshold to obtain an isolated forest optimization model, and performing temperature anomaly analysis on the preheating pipeline by using the isolated forest optimization model;
the step of obtaining the fluctuation factor according to the data fluctuation characteristics in the preset neighborhood of the local minimum point in the local extremum curve comprises the following steps:
calculating the absolute value of the slope of a straight line between a local extremum point and an adjacent local extremum point in the local extremum curve, and obtaining the slope representation value of the local extremum point; calculating the variance of the slope representation value of the local extreme point in the preset neighborhood of the local minimum point to obtain the fluctuation factor of the local minimum point;
the step of obtaining the difference factor according to the difference characteristics of the local minimum point and other local extremum points in the local extremum curve comprises the following steps:
calculating the numerical average value of local extremum points in the local extremum curve to obtain an integral characterization value; calculating the numerical value difference between the integral characterization value and the local minimum value point to obtain an integral difference value; when the integral difference value is a positive number, calculating the ratio of the integral difference value to the integral characterization value to obtain a difference factor of the local minimum point; when the integral difference value is not positive, the difference factor of the local minimum value point is a preset first constant;
the step of obtaining the change factor according to the discrete characteristics of the local minimum point and the adjacent local extremum point in the local extremum curve comprises the following steps:
calculating the average value of the slope characterization values of the local minimum points and the adjacent local extremum points to obtain a change factor of the local minimum;
the step of obtaining the recommended degree of the local minimum point according to the fluctuation factor, the difference factor and the change factor comprises the following steps:
and calculating the product of the fluctuation factor, the difference factor and the change factor of the local minimum point to obtain the recommended degree of the local minimum point.
2. The method for analyzing the abnormality of the temperature data of the preheating pipeline for aluminum metal smelting according to claim 1, wherein the step of acquiring the abnormality distribution curve of the abnormality fraction of the temperature monitoring dataset according to the isolated forest model comprises:
obtaining abnormal fractions of each temperature monitoring sample in the temperature monitoring data set according to the isolated forest model; and constructing a two-dimensional rectangular coordinate system of an abnormal distribution curve according to the abnormal scores of the temperature monitoring samples, wherein the transverse axis of the two-dimensional rectangular coordinate system of the abnormal distribution curve is the abnormal scores sequenced from small to large, the vertical axis is the number value of the temperature monitoring samples corresponding to the abnormal scores, and connecting the number values of the temperature monitoring samples corresponding to each abnormal distribution to obtain the abnormal distribution curve.
3. The method for analyzing abnormal temperature data of a preheating pipeline for aluminum metal smelting according to claim 1, wherein the step of obtaining a local extremum curve according to the data change characteristics of the abnormal distribution curve comprises the steps of:
and obtaining local extremum points in the abnormal distribution curve, constructing a two-dimensional rectangular coordinate system related to the local extremum curve, wherein the horizontal axis of the two-dimensional rectangular coordinate system of the local extremum curve is an abnormal fraction corresponding to the local extremum points, the vertical axis is a numerical value of the local extremum points, and connecting the numerical values of each local extremum point to obtain the local extremum curve.
4. The method for analyzing abnormal temperature data of a preheating pipeline for aluminum metal smelting according to claim 1, wherein the step of obtaining the recommended threshold value of the isolated forest model according to the recommended degree of all local minimum points comprises the steps of:
and taking the abnormal score corresponding to the local minimum point of the recommended degree maximum value in the local extremum curve as a recommended threshold value of the isolated forest model.
5. The method for analyzing abnormal temperature data of a preheating pipeline for aluminum metal smelting according to claim 1, wherein the step of obtaining the detection accuracy of an isolated forest model according to the recommended threshold value comprises the steps of:
obtaining abnormal samples in the temperature monitoring data set through an isolated forest model according to the recommended threshold value, and counting the number of actual abnormal samples in the abnormal samples to obtain abnormal quantity; and calculating the ratio of the abnormal quantity to the number of the abnormal samples to obtain the detection accuracy of the isolated forest model.
6. The method for analyzing abnormal temperature data of a preheating pipeline for aluminum metal smelting according to claim 1, wherein the step of obtaining the optimal recommended threshold of the isolated forest model according to all recommended thresholds and detection accuracy comprises the steps of:
calculating the average value of the absolute values of the differences between the recommendation threshold and other recommendation thresholds, and obtaining a central characterization value of the recommendation threshold; taking the recommended threshold with the maximum central characterization value as a reference recommended threshold; calculating the absolute value of the difference between any recommended threshold and a reference recommended threshold and carrying out negative correlation mapping to obtain a distance characterization value of the any recommended threshold; and calculating multiplication results among the arbitrary recommendation threshold, the distance characterization value and the detection accuracy aiming at the arbitrary recommendation threshold, and taking the average value of the multiplication results corresponding to all recommendation thresholds as the optimal recommendation threshold of the isolated forest model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410043822.1A CN117556714B (en) | 2024-01-12 | 2024-01-12 | Preheating pipeline temperature data anomaly analysis method for aluminum metal smelting |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410043822.1A CN117556714B (en) | 2024-01-12 | 2024-01-12 | Preheating pipeline temperature data anomaly analysis method for aluminum metal smelting |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117556714A CN117556714A (en) | 2024-02-13 |
| CN117556714B true CN117556714B (en) | 2024-03-22 |
Family
ID=89813306
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410043822.1A Active CN117556714B (en) | 2024-01-12 | 2024-01-12 | Preheating pipeline temperature data anomaly analysis method for aluminum metal smelting |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117556714B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117786371B (en) * | 2024-02-27 | 2024-05-10 | 聊城市检验检测中心 | Temperature monitoring data optimization prediction analysis method and system |
| CN117879769B (en) * | 2024-03-11 | 2024-05-28 | 陕西惠延机械有限公司 | Data pushing and transmitting method for tunnel lining trolley cloud platform system |
| CN117951616B (en) * | 2024-03-27 | 2024-05-28 | 山东海纳智能装备科技股份有限公司 | Intelligent operation and maintenance analysis method for local ventilator |
| CN118009914B (en) * | 2024-04-08 | 2024-06-11 | 上海中医药大学附属岳阳中西医结合医院 | Infrared spectrum-based intelligent moxibustion robot part temperature deformation monitoring method |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111008662A (en) * | 2019-12-04 | 2020-04-14 | 贵州电网有限责任公司 | Online monitoring data anomaly analysis method for power transmission line |
| WO2022142042A1 (en) * | 2020-12-29 | 2022-07-07 | 平安科技（深圳）有限公司 | Abnormal data detection method and apparatus, computer device and storage medium |
| CN115577275A (en) * | 2022-11-11 | 2023-01-06 | 山东产业技术研究院智能计算研究院 | Time sequence data anomaly monitoring system and method based on LOF and isolated forest |
| CN116011894A (en) * | 2023-03-28 | 2023-04-25 | 河北长发铝业股份有限公司 | Aluminum alloy rod production data management system |
| CN117195139A (en) * | 2023-11-08 | 2023-12-08 | 北京珺安惠尔健康科技有限公司 | Chronic disease health data dynamic monitoring method based on machine learning |
| CN117349764A (en) * | 2023-12-05 | 2024-01-05 | 河北三臧生物科技有限公司 | Intelligent analysis method for stem cell induction data |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108777873B (en) * | 2018-06-04 | 2021-03-02 | 江南大学 | Wireless sensor network abnormal data detection method based on weighted mixed isolated forest |
-
2024
- 2024-01-12 CN CN202410043822.1A patent/CN117556714B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111008662A (en) * | 2019-12-04 | 2020-04-14 | 贵州电网有限责任公司 | Online monitoring data anomaly analysis method for power transmission line |
| WO2022142042A1 (en) * | 2020-12-29 | 2022-07-07 | 平安科技（深圳）有限公司 | Abnormal data detection method and apparatus, computer device and storage medium |
| CN115577275A (en) * | 2022-11-11 | 2023-01-06 | 山东产业技术研究院智能计算研究院 | Time sequence data anomaly monitoring system and method based on LOF and isolated forest |
| CN116011894A (en) * | 2023-03-28 | 2023-04-25 | 河北长发铝业股份有限公司 | Aluminum alloy rod production data management system |
| CN117195139A (en) * | 2023-11-08 | 2023-12-08 | 北京珺安惠尔健康科技有限公司 | Chronic disease health data dynamic monitoring method based on machine learning |
| CN117349764A (en) * | 2023-12-05 | 2024-01-05 | 河北三臧生物科技有限公司 | Intelligent analysis method for stem cell induction data |
Non-Patent Citations (3)
| Title |
|---|
| Application of Isolated Forest Algorithm in Deep Learning Change Detection of High Resolution Remote Sensing Image;Wenchun Zhang.etc;《IEEE》;20201231;全文 * |
| 基于孤立森林算法的台区线损分析与管理***研究;周昕;张怡;王桢干;唐恬;;电力学报;20191225(06);全文 * |
| 孤立森林算法在大坝监测数据异常识别中的应用;张海龙;范振东;陈敏;;人民黄河;20200810(08);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117556714A (en) | 2024-02-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN117556714B (en) | Preheating pipeline temperature data anomaly analysis method for aluminum metal smelting | |
| CN110427996B (en) | Fuzzy matching-based time sequence abnormal pattern recognition method and device | |
| CN115982602B (en) | Photovoltaic transformer electrical fault detection method | |
| CN107179310B (en) | Raman spectrum characteristic peak recognition methods based on robust noise variance evaluation | |
| CN111275307A (en) | Quality control method for high-frequency continuous observation data of automatic online water quality station | |
| CN109446189A (en) | A kind of technological parameter outlier detection system and method | |
| CN116881745B (en) | Pressure transmitter abnormality monitoring method based on big data | |
| CN116756595B (en) | Conductive slip ring fault data acquisition and monitoring method | |
| CN115858894B (en) | Visual big data analysis method | |
| CN112380992B (en) | Method and device for evaluating and optimizing accuracy of monitoring data in machining process | |
| CN110889441A (en) | Distance and point density based substation equipment data anomaly identification method | |
| CN116610938B (en) | Method and equipment for detecting unsupervised abnormality of semiconductor manufacture in curve mode segmentation | |
| CN113757093A (en) | Fault diagnosis method for flash steam compressor unit | |
| CN113536066A (en) | Data anomaly detection algorithm determination method and device and computer equipment | |
| CN114297264B (en) | Method and system for detecting abnormal fragments of time sequence signals | |
| CN114817228A (en) | Abnormal value removing method for long-time sequence monitoring data of lake water level | |
| CN109844779B (en) | Method and system for analyzing measurement-yield correlation | |
| CN117491422B (en) | Method for detecting high heat-conducting property of aluminum alloy material | |
| CN117517596B (en) | Method and system for monitoring combustible and toxic harmful gases in real time based on Internet of things | |
| CN117557566B (en) | Intelligent detection method for quality of heating wire of electronic cigarette based on image characteristics | |
| CN117540325B (en) | Business database anomaly detection method and system based on data variation capture | |
| CN117314020A (en) | Wetland carbon sink data monitoring system of plankton | |
| CN109163894B (en) | Running-in state identification method based on friction temperature signal | |
| CN115628776A (en) | Water supply pipe network abnormal data detection method | |
| CN112037158B (en) | Shale gas field production equipment-based image enhancement labeling method |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |