CN117034083A - Method, device and equipment for selecting interpretation method of partial discharge classification model - Google Patents
Method, device and equipment for selecting interpretation method of partial discharge classification model Download PDFInfo
- Publication number
- CN117034083A CN117034083A CN202311024710.3A CN202311024710A CN117034083A CN 117034083 A CN117034083 A CN 117034083A CN 202311024710 A CN202311024710 A CN 202311024710A CN 117034083 A CN117034083 A CN 117034083A
- Authority
- CN
- China
- Prior art keywords
- partial discharge
- discharge classification
- result
- interpretability
- interpretation
- 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
- 238000000034 method Methods 0.000 title claims abstract description 184
- 238000013145 classification model Methods 0.000 title claims abstract description 175
- 238000013528 artificial neural network Methods 0.000 claims abstract description 18
- 230000015654 memory Effects 0.000 claims description 26
- 238000004590 computer program Methods 0.000 claims description 22
- 238000012163 sequencing technique Methods 0.000 claims description 20
- 238000000605 extraction Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims 2
- 238000012545 processing Methods 0.000 description 7
- 238000013527 convolutional neural network Methods 0.000 description 6
- 230000007547 defect Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 238000011156 evaluation Methods 0.000 description 4
- 238000013473 artificial intelligence Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000009413 insulation Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000009466 transformation Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000013500 data storage Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 1
- 238000003646 Spearman's rank correlation coefficient Methods 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 230000005684 electric field Effects 0.000 description 1
- 238000007667 floating Methods 0.000 description 1
- 229910021389 graphene Inorganic materials 0.000 description 1
- 239000011810 insulating material Substances 0.000 description 1
- 239000012212 insulator Substances 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 238000010187 selection method Methods 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
-
- 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
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Economics (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application relates to a method, a device and equipment for selecting an explanatory method of a partial discharge classification model. The method comprises the following steps: obtaining partial discharge classification models under different neural network architectures, and explaining various partial discharge classification models by utilizing various interpretable methods to obtain interpretation information corresponding to each interpretable method; acquiring a preset interpretability result corresponding to each partial discharge classification model, and comparing the interpretation information corresponding to the interpretability method with the preset interpretability results corresponding to the various partial discharge classification models for each interpretability method to acquire a comparison result; a target one of the plurality of interpretive methods is selected based on the comparison result. The method can be used for rapidly determining the explanatory method of the partial discharge classification model.
Description
Technical Field
The present application relates to the field of partial discharge technologies, and in particular, to a method, an apparatus, and a device for selecting an interpretable method of a partial discharge classification model.
Background
With the rapid development of the electric power industry and the high-speed expansion of the electric network, more and more new technologies are required to support and maintain the operation safety of the electric network. The safe and stable operation of the power grid is the basis of reliable power supply, once the power grid fails, the power equipment is damaged, the power supply is stopped, normal production and life of people are affected, public safety is also jeopardized in severe cases, and serious economic loss and bad social influence are caused.
Partial discharge generally refers to a discharge process generated by concentration of a partial electric field in an insulator or on an insulating surface of power equipment, and in the discharge process, phenomena such as sound, light, heat and chemical reaction are accompanied, and under the action of partial discharge, the aging of insulating materials of the power equipment is aggravated, and finally insulation breakdown is caused to fail, so that detection of a partial discharge signal is a key of evaluation of the insulation state and fault positioning of the power equipment, and is directly related to establishment of overhaul and maintenance schemes of the power equipment, so that stable operation of a power system is affected, the partial discharge signal is classified, and the insulation condition of the power equipment on site can be diagnosed according to the acquired partial discharge signal.
In order to improve the diagnosis efficiency and accuracy of the partial discharge fault type, machine learning and artificial intelligence technology are currently used for online detection and evaluation of power transformation equipment. In the context of rapid development and application of artificial intelligence, it becomes critical to interpret the results of the algorithm output to the user, known as the post-mortem method. The decision made by the model can be understood by researching a post-interpretation method, and the cause of deviation is found out, so that the performance of the model is improved; the method can also help the user to understand the decision made by the artificial intelligence, improve the reliability of the result and enable the user to use the model more effectively. However, the current interpretability methods are numerous and wide in range, and different interpretability methods are suitable for different scenes, so that it is difficult to select an appropriate interpretability method for interpreting the partial discharge classification model.
Disclosure of Invention
In view of the foregoing, it is desirable to provide a method, apparatus and device for selecting a partial discharge classification model interpretability method that can quickly determine the partial discharge classification model interpretability method.
In a first aspect, the present application provides a method of selecting an interpretation method for a classification model of partial discharge. The method comprises the following steps:
the method comprises the steps of obtaining partial discharge classification models under different neural network architectures, and utilizing a plurality of interpretable methods to interpret various partial discharge classification models to obtain interpretation information corresponding to each interpretable method, wherein the interpretation information is used for representing the contribution degree of each attribute characteristic in various partial discharge classification models to partial discharge classification results;
acquiring preset interpretability results corresponding to each partial discharge classification model, and comparing interpretation information corresponding to the interpretability methods with the preset interpretability results corresponding to various partial discharge classification models for each interpretability method to acquire comparison results;
according to the comparison result, a target interpretability method of the plurality of interpretability methods is selected.
In one embodiment, obtaining a preset interpretability result corresponding to each partial discharge classification model includes:
carrying out feature extraction on partial discharge ultrahigh frequency data under different fault types to obtain initial feature data of various attribute features;
acquiring the association degree of each attribute feature and the fault type based on initial feature data under different fault types, and acquiring at least one target attribute feature in each attribute feature based on the association degree;
and for each target attribute feature, disturbing feature data corresponding to the target attribute feature in the initial feature data to obtain disturbance feature data corresponding to the target attribute feature, and determining a preset interpretability result corresponding to each partial discharge classification model by combining the initial feature data and the disturbance feature data.
In one embodiment, determining the preset interpretability result corresponding to each partial discharge classification model by combining the initial feature data and the disturbance feature data includes:
combining the initial characteristic data, the disturbance characteristic data and the partial discharge classification model, and determining the target contribution degree of the target attribute characteristic to the partial discharge classification model;
aiming at each partial discharge classification model, sequencing the target contribution degree of each target attribute feature to the partial discharge classification model to obtain a target sequencing result of each target attribute feature in the partial discharge classification model, wherein the target sequencing result is used for representing a preset interpretability result corresponding to the partial discharge classification model.
In one embodiment, comparing the interpretation information corresponding to the interpretation method with preset interpretation results corresponding to various partial discharge classification models to obtain comparison results, including:
comparing interpretation sub-information corresponding to the partial discharge classification model in the interpretation information with a preset interpretability result aiming at each partial discharge classification model to obtain a comparison sub-result, wherein the interpretation sub-information is used for representing the contribution degree of each attribute feature in one partial discharge classification model to the partial discharge classification result;
and taking the average value of the comparison sub-results corresponding to the partial discharge classification models in the interpretation information of the interpretability method to obtain the comparison result of the interpretability method.
In one embodiment, comparing interpretation sub-information corresponding to the partial discharge classification model in the interpretation information with a preset interpretability result to obtain a comparison sub-result, including:
obtaining the similarity degree of the interpretation sub-information and a preset interpretation result;
and obtaining a comparison sub-result according to the similarity degree.
In one embodiment, obtaining the similarity degree between the interpretation sub-information and the preset interpretable result includes:
judging the sequencing result of each target attribute feature in the interpretation sub-information according to the contribution degree of each attribute feature in the interpretation sub-information to the partial discharge classification result;
and obtaining the similarity according to the target sequencing result in the sequencing result and the preset interpretability result.
In a second aspect, the application further provides a device for selecting the interpretation method of the partial discharge classification model. The device comprises:
the acquisition module is used for acquiring partial discharge classification models under different neural network architectures, and utilizing various interpretable methods to interpret various partial discharge classification models to obtain interpretation information corresponding to each interpretable method, wherein the interpretation information is used for representing the contribution degree of each attribute characteristic in various partial discharge classification models to the partial discharge classification result;
the comparison module is used for acquiring preset interpretability results corresponding to each partial discharge classification model, and comparing interpretation information corresponding to the interpretability methods with the preset interpretability results corresponding to the various partial discharge classification models for each interpretability method to acquire comparison results;
and the selection module is used for selecting a target interpretability method in the plurality of interpretability methods according to the comparison result.
In a third aspect, the present application also provides a computer device. The computer device comprises a memory and a processor, the memory stores a computer program, and the processor executes the computer program to realize the following steps:
the method comprises the steps of obtaining partial discharge classification models under different neural network architectures, and utilizing a plurality of interpretable methods to interpret various partial discharge classification models to obtain interpretation information corresponding to each interpretable method, wherein the interpretation information is used for representing the contribution degree of each attribute characteristic in various partial discharge classification models to partial discharge classification results;
acquiring preset interpretability results corresponding to each partial discharge classification model, and comparing interpretation information corresponding to the interpretability methods with the preset interpretability results corresponding to various partial discharge classification models for each interpretability method to acquire comparison results;
according to the comparison result, a target interpretability method of the plurality of interpretability methods is selected.
In a fourth aspect, the present application also provides a computer-readable storage medium. The computer readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of:
the method comprises the steps of obtaining partial discharge classification models under different neural network architectures, and utilizing a plurality of interpretable methods to interpret various partial discharge classification models to obtain interpretation information corresponding to each interpretable method, wherein the interpretation information is used for representing the contribution degree of each attribute characteristic in various partial discharge classification models to partial discharge classification results;
acquiring preset interpretability results corresponding to each partial discharge classification model, and comparing interpretation information corresponding to the interpretability methods with the preset interpretability results corresponding to various partial discharge classification models for each interpretability method to acquire comparison results;
according to the comparison result, a target interpretability method of the plurality of interpretability methods is selected.
In a fifth aspect, the present application also provides a computer program product. The computer program product comprising a computer program which, when executed by a processor, performs the steps of:
the method comprises the steps of obtaining partial discharge classification models under different neural network architectures, and utilizing a plurality of interpretable methods to interpret various partial discharge classification models to obtain interpretation information corresponding to each interpretable method, wherein the interpretation information is used for representing the contribution degree of each attribute characteristic in various partial discharge classification models to partial discharge classification results;
acquiring preset interpretability results corresponding to each partial discharge classification model, and comparing interpretation information corresponding to the interpretability methods with the preset interpretability results corresponding to various partial discharge classification models for each interpretability method to acquire comparison results;
according to the comparison result, a target interpretability method of the plurality of interpretability methods is selected.
The method, the device and the equipment for selecting the partial discharge classification model interpretability method acquire partial discharge classification models under different neural network architectures, interpret various partial discharge classification models by utilizing various interpretability methods to acquire interpretation information corresponding to each type of the partial discharge classification models, acquire preset interpretability results corresponding to each type of the partial discharge classification models, compare the interpretation information corresponding to the interpretability methods with the preset interpretability results corresponding to the various types of the partial discharge classification models for each type of the partial discharge classification models, acquire comparison results, and select a target interpretability method in the various types of the interpretability methods according to the comparison results. According to the application, the local discharge classification models with different architectures are selected, the multiple kinds of interpretable methods are applied to the local discharge classification models with different architectures, the interpretation of each type of interpretable method for each local discharge classification model is obtained, the interpretation information of each type of interpretable method is obtained, the similarity degree or the difference degree of the interpretation information of each type of interpretable method and the preset interpretable result is obtained by comparing the interpretation information with the preset interpretable result, the interpretable method which is more similar to the preset interpretable result can be selected from the local discharge classification models with different architectures, the robustness and the stability are improved, the target interpretable method which is more suitable for the local discharge classification model can be selected from the multiple kinds of interpretable methods, and the interpretation evaluation of other local discharge classification models can be carried out rapidly by using the target interpretable method.
Drawings
FIG. 1 is a diagram of an application environment for a method of selecting an explanatory method of a partial discharge classification model in one embodiment;
FIG. 2 is a flow diagram of a method of selecting an explanatory method of a partial discharge classification model in one embodiment;
FIG. 3 is a flow chart of determining a preset interpretability result in one embodiment;
FIG. 4 is a flow chart of a comparison result in one embodiment;
FIG. 5 is a block diagram of a selection device of an explanatory method of partial discharge classification model in one embodiment;
fig. 6 is an internal structural diagram of a computer device in one embodiment.
Detailed Description
The present application will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present application more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the application.
The method for selecting the explanatory method of the partial discharge classification model provided by the embodiment of the application can be applied to an application environment shown in fig. 1. Wherein the terminal 102 communicates with the server 104 via a network. The data storage system may store data that the server 104 needs to process. The data storage system may be integrated on the server 104 or may be located on a cloud or other network server. The terminal 102 may be, but not limited to, various personal computers, notebook computers, smart phones, tablet computers, internet of things devices, and portable wearable devices, where the internet of things devices may be smart speakers, smart televisions, smart air conditioners, smart vehicle devices, and the like. The portable wearable device may be a smart watch, smart bracelet, headset, or the like. The server 104 may be implemented as a stand-alone server or as a server cluster of multiple servers.
In one embodiment, as shown in fig. 2, a method for selecting an explanatory method of a classification model of partial discharge is provided, and the method is applied to the server in fig. 1 for illustration, and includes the following steps:
step 202, obtaining partial discharge classification models under different neural network architectures, and utilizing a plurality of interpretable methods to interpret various partial discharge classification models to obtain interpretation information corresponding to each interpretable method.
The interpretation information is used for representing the contribution degree of each attribute characteristic in various partial discharge classification models to the partial discharge classification result.
The embodiment of the application preferably obtains partial discharge classification models under different neural network architectures, such as a Convolutional Neural Network (CNN) partial discharge classification model based on a CNN, a Bi-LSTM partial discharge classification model based on a Bi-LSTM based on a Bi-long short-term memory network (Bi-LSTM), a transducer partial discharge classification model based on a transducer neural network, and the like.
For example, the uhf signal data sets under different partial discharge fault type labels, for example, the uhf signal data sets under different fault types such as a floating potential defect, a tip defect, an air gap discharge defect, a creeping discharge defect, and a particle discharge defect, may be first acquired. Basic model architectures of a convolutional neural network, a two-way long-short-term memory network and a Transformer neural network are built, and the basic model architectures are trained through ultrahigh frequency signal data sets under different fault types, so that partial discharge classification models under different neural network architectures are obtained.
The embodiment of the application utilizes a plurality of interpretable methods to interpret various partial discharge classification models, and obtains interpretation information corresponding to each interpretable method, wherein the interpretation information comprises the interpretation of each interpretable method for each partial discharge classification model. The interpretability approach may identify attribute features that have an impact on model predictions, such as analyzing weights of model parameters, contribution of features, or other relevant metrics to quantify the importance of features. The interpretation information of the embodiment of the application is used for representing the contribution degree of each attribute feature in various partial discharge classification models to the partial discharge classification result, and each attribute feature is a feature extracted from the ultrahigh frequency signal data set.
Illustratively, six interpretable methods of DeepLift, gradShap, integrated Gradients, kernelShap, deepLiftShap and Shapley are chosen for application in the three neural network architectures described above. Taking the deep method as an example, the interpretation information corresponding to the deep method includes interpretation of the CNN partial discharge classification model, interpretation of the Bi-LSTM partial discharge classification model and interpretation of the transducer partial discharge classification model by the deep method, namely the contribution degree of each attribute feature in the CNN partial discharge classification model interpreted by the deep method to the partial discharge classification result, the contribution degree of each attribute feature in the Bi-LSTM partial discharge classification model to the partial discharge classification result and the contribution degree of each attribute feature in the transducer partial discharge classification model to the partial discharge classification result.
Step 204, obtaining a preset interpretability result corresponding to each partial discharge classification model, and comparing the interpretation information corresponding to the interpretability method with the preset interpretability results corresponding to the various partial discharge classification models for each interpretability method to obtain a comparison result.
The embodiment of the application obtains the corresponding preset interpretability result of each partial discharge classification model, and compares the interpretation of each partial discharge classification model in each interpretability method with the corresponding preset interpretability result to obtain the comparison result of each interpretability method. The comparison result may be the similarity degree or the difference degree between the interpretation information corresponding to the interpretable method and the corresponding preset interpretable result.
Step 206, selecting a target interpretability method of the plurality of interpretability methods according to the comparison result.
According to the comparison result, the embodiment of the application selects the interpretable method with high similarity or small difference with the preset interpretable result as the target interpretable method.
In the method for selecting the partial discharge classification model interpretability method, the partial discharge classification models under different neural network architectures are obtained, various partial discharge classification models are interpreted by utilizing various interpretable methods, interpretation information corresponding to each type of the interpretable methods is obtained, a preset interpretable result corresponding to each type of the partial discharge classification models is obtained, the interpretation information corresponding to each type of the interpretable methods is compared with the preset interpretable results corresponding to the various types of the partial discharge classification models for each type of the interpretable methods, a comparison result is obtained, and a target interpretable method in the various types of the interpretable methods is selected according to the comparison result. According to the embodiment of the application, the local discharge classification models with different architectures are selected, the multiple kinds of interpretable methods are applied to the local discharge classification models with different architectures, the interpretation of each type of interpretable method for each local discharge classification model is obtained, the interpretation information of each type of interpretable method is obtained, the similarity degree or the difference degree between the interpretation information of each type of interpretable method and the preset interpretable result is obtained by comparing the interpretation information with the preset interpretable result, the interpretable method which is more similar to the preset interpretable result can be selected from the local discharge classification models with different architectures, the robustness and the stability are improved, the target interpretable method of the local discharge classification model which is more suitable can be selected from the multiple kinds of interpretable methods, and the interpretation evaluation of other local discharge classification models can be rapidly carried out by using the target interpretable method.
In one embodiment, as shown in fig. 3, obtaining the preset interpretability result corresponding to each partial discharge classification model includes:
and 302, carrying out feature extraction on the partial discharge ultrahigh frequency data under different fault types to obtain initial feature data of various attribute features.
The embodiment of the application can firstly acquire partial discharge original data carrying different fault type labels, such as partial discharge ultrahigh frequency data, wherein the partial discharge ultrahigh frequency data is acquired time sequence signal data. And extracting the characteristics of the partial discharge ultrahigh frequency data to obtain initial characteristic data of various attribute characteristics.
Step 304, based on the initial feature data under different fault types, obtaining the association degree of each attribute feature and the fault type, and based on the association degree, obtaining at least one target attribute feature in each attribute feature.
After the feature data of various attribute features under different fault types are obtained, the embodiment of the application can obtain the association degree of each attribute feature and the fault type according to the feature data of each attribute feature and the corresponding fault type label, and the association degree can reflect the correlation coefficient of each attribute feature and the fault type label in the feature data.
After the relevance is determined, the embodiment of the application obtains at least one target attribute feature in all attribute features. For example, the size of the association degree and the preset threshold is judged, and the attribute features corresponding to the association degree larger than the preset threshold are used as target attribute features. For another example, the relevance of each attribute feature and the fault type is ranked, and a preset number of attribute features with higher relevance are selected as target attribute features.
Step 306, for each target attribute feature, perturbation is performed on feature data corresponding to the target attribute feature in the initial feature data, perturbation feature data corresponding to the target attribute feature is obtained, and the initial feature data and the perturbation feature data are combined to determine a preset interpretability result corresponding to each partial discharge classification model.
According to the embodiment of the application, for each target attribute feature, the feature data corresponding to the target attribute feature in the initial feature data is disturbed, and disturbance feature data corresponding to the target attribute feature is obtained.
For example, noise processing or transformation processing is performed on feature data corresponding to the target attribute feature 1 in the initial feature data, so as to obtain disturbance feature data corresponding to the target attribute feature 1.
For another example, noise processing or transformation processing is performed on feature data corresponding to the target attribute feature 2 in the initial feature data, so as to obtain disturbance feature data corresponding to the target attribute feature 2.
The disturbance characteristic data of the embodiment of the application is only disturbance aiming at a certain target attribute characteristic on the basis of the initial characteristic data, so that the target contribution degree of the target attribute characteristic to the classification result of the partial discharge classification model can be determined by combining the initial characteristic data, the disturbance characteristic data and the partial discharge classification model.
Specifically, the initial characteristic data, the disturbance characteristic data and the partial discharge classification models are combined, the target contribution degree of each target attribute characteristic to each partial discharge classification model is determined, the target contribution degree of each target attribute characteristic to the partial discharge classification model is ranked according to the target contribution degree of each target attribute characteristic to each partial discharge classification model, and a target ranking result of each target attribute characteristic in the partial discharge classification model is obtained, wherein the target ranking result is used for representing a preset interpretability result corresponding to the partial discharge classification model.
In one embodiment, as shown in fig. 4, comparing the interpretation information corresponding to the interpretation method with preset interpretation results corresponding to various partial discharge classification models to obtain comparison results, including:
step 402, for each partial discharge classification model, comparing the interpretation sub-information corresponding to the partial discharge classification model in the interpretation information with a preset interpretation result to obtain a comparison sub-result.
The interpretation sub-information is used for representing the contribution degree of each attribute characteristic in a partial discharge classification model to the partial discharge classification result.
The interpretation information corresponding to each interpretable method of the embodiment of the application comprises interpretation sub-information of each partial discharge classification model. And comparing the interpretation sub-information with the corresponding preset interpretable result to obtain a comparison sub-result.
For example, the similarity degree of the interpretation sub-information and the preset interpretable result may be obtained, and the comparison sub-result may be obtained by the similarity degree of the interpretation sub-information and the preset interpretable result. For example, according to the contribution degree of each attribute feature in the interpretation sub-information to the partial discharge classification result, judging the sorting result of each target attribute feature in the interpretation sub-information; and obtaining the similarity according to the target sequencing result in the sequencing result and the preset interpretability result.
For example, a consistent pair and an inconsistent pair between the sorting result of each target attribute feature and the target sorting result in the interpretation sub-information may be acquired, so that the correlation coefficient is calculated to determine the degree of similarity, such as a kendel rank correlation coefficient, a spearman rank correlation coefficient, or the like.
And step 404, taking the average value of the comparison sub-results corresponding to the partial discharge classification models in the interpretation information of the interpretability method to obtain the comparison result of the interpretability method.
After each comparison sub-result in the interpretation information of each interpretable method is obtained, the embodiment of the application takes the average value of each comparison sub-result, thereby being capable of representing the interpretation performance of the interpretable method on the partial discharge classification model under different neural network architectures.
It should be understood that, although the steps in the flowcharts related to the embodiments described above are sequentially shown as indicated by arrows, these steps are not necessarily sequentially performed in the order indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps in the flowcharts described in the above embodiments may include a plurality of steps or a plurality of stages, which are not necessarily performed at the same time, but may be performed at different times, and the order of the steps or stages is not necessarily performed sequentially, but may be performed alternately or alternately with at least some of the other steps or stages.
Based on the same inventive concept, the embodiment of the application also provides a device for selecting the partial discharge classification model interpretability method for realizing the method for selecting the partial discharge classification model interpretability method. The implementation of the solution provided by the device is similar to the implementation described in the above method, so the specific limitation in the embodiments of the selection device of the one or more partial discharge classification model interpretability methods provided below may be referred to the limitation of the selection method of the partial discharge classification model interpretability method hereinabove, and will not be repeated here.
In one embodiment, as shown in fig. 5, there is provided a selection apparatus of an interpretable method of a partial discharge classification model, including: an acquisition module 502, a comparison module 504, and a selection module 506, wherein:
the obtaining module 502 is configured to obtain partial discharge classification models under different neural network architectures, and interpret various partial discharge classification models by using various interpretable methods to obtain interpretation information corresponding to each interpretable method, where the interpretation information is used to characterize contribution degrees of each attribute feature in the various partial discharge classification models to the partial discharge classification result.
The comparison module 504 is configured to obtain preset interpretability results corresponding to each partial discharge classification model, and compare, for each interpretability method, interpretation information corresponding to the interpretability method with preset interpretability results corresponding to various partial discharge classification models, thereby obtaining comparison results.
A selection module 506, configured to select a target interpretability method of the plurality of interpretability methods according to the comparison result.
In one embodiment, the comparison module 504, when executing the obtaining of the preset interpretability results corresponding to each partial discharge classification model, is configured to: carrying out feature extraction on partial discharge ultrahigh frequency data under different fault types to obtain initial feature data of various attribute features; acquiring the association degree of each attribute feature and the fault type based on initial feature data under different fault types, and acquiring at least one target attribute feature in each attribute feature based on the association degree; and for each target attribute feature, disturbing feature data corresponding to the target attribute feature in the initial feature data to obtain disturbance feature data corresponding to the target attribute feature, and determining a preset interpretability result corresponding to each partial discharge classification model by combining the initial feature data and the disturbance feature data.
In one embodiment, the comparison module 504, when executing the determination of the preset interpretability result corresponding to each partial discharge classification model in combination with the initial feature data and the disturbance feature data, is configured to: combining the initial characteristic data, the disturbance characteristic data and the partial discharge classification model, and determining the target contribution degree of the target attribute characteristic to the partial discharge classification model; aiming at each partial discharge classification model, sequencing the target contribution degree of each target attribute feature to the partial discharge classification model to obtain a target sequencing result of each target attribute feature in the partial discharge classification model, wherein the target sequencing result is used for representing a preset interpretability result corresponding to the partial discharge classification model.
In one embodiment, the comparison module 504 performs comparison of the interpretation information corresponding to the interpretation method with preset interpretation results corresponding to various partial discharge classification models, and when obtaining the comparison result, is used for: comparing interpretation sub-information corresponding to the partial discharge classification model in the interpretation information with a preset interpretability result aiming at each partial discharge classification model to obtain a comparison sub-result, wherein the interpretation sub-information is used for representing the contribution degree of each attribute feature in one partial discharge classification model to the partial discharge classification result; and taking the average value of the comparison sub-results corresponding to the partial discharge classification models in the interpretation information of the interpretability method to obtain the comparison result of the interpretability method.
In one embodiment, the comparing module 504 is configured to compare the interpretation sub-information corresponding to the partial discharge classification model in the interpretation information with a preset interpretation result, and when obtaining the comparison sub-result, is configured to: obtaining the similarity degree of the interpretation sub-information and a preset interpretation result; and obtaining a comparison sub-result according to the similarity degree.
In one embodiment, the comparing module 504 is configured to, when executing the obtaining of the similarity between the interpretation sub-information and the preset interpretability result: judging the sequencing result of each target attribute feature in the interpretation sub-information according to the contribution degree of each attribute feature in the interpretation sub-information to the partial discharge classification result; and obtaining the similarity according to the target sequencing result in the sequencing result and the preset interpretability result.
The respective modules in the above-described selection means of the partial discharge classification model interpretability method may be implemented in whole or in part by software, hardware, and combinations thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.
In one embodiment, a computer device is provided, which may be a server, the internal structure of which may be as shown in fig. 6. The computer device includes a processor, a memory, an Input/Output interface (I/O) and a communication interface. The processor, the memory and the input/output interface are connected through a system bus, and the communication interface is connected to the system bus through the input/output interface. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database of the computer device is used for storing partial discharge data. The input/output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a method of selecting a partial discharge classification model interpretability method.
It will be appreciated by those skilled in the art that the structure shown in FIG. 6 is merely a block diagram of some of the structures associated with the present inventive arrangements and is not limiting of the computer device to which the present inventive arrangements may be applied, and that a particular computer device may include more or fewer components than shown, or may combine some of the components, or have a different arrangement of components.
In one embodiment, a computer device is provided, comprising a memory and a processor, the memory having stored therein a computer program, the processor implementing the steps of the method embodiments described above when the computer program is executed.
In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored which, when executed by a processor, implements the steps of the method embodiments described above.
In an embodiment, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the steps of the method embodiments described above.
It should be noted that, the user information (including but not limited to user equipment information, user personal information, etc.) and the data (including but not limited to data for analysis, stored data, presented data, etc.) related to the present application are information and data authorized by the user or sufficiently authorized by each party, and the collection, use and processing of the related data need to comply with the related laws and regulations and standards of the related country and region.
Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, database, or other medium used in embodiments provided herein may include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, high density embedded nonvolatile Memory, resistive random access Memory (ReRAM), magnetic random access Memory (Magnetoresistive Random Access Memory, MRAM), ferroelectric Memory (Ferroelectric Random Access Memory, FRAM), phase change Memory (Phase Change Memory, PCM), graphene Memory, and the like. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory, and the like. By way of illustration, and not limitation, RAM can be in the form of a variety of forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), and the like. The databases referred to in the embodiments provided herein may include at least one of a relational database and a non-relational database. The non-relational database may include, but is not limited to, a blockchain-based distributed database, and the like. The processor referred to in the embodiments provided in the present application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic unit, a data processing logic unit based on quantum computing, or the like, but is not limited thereto.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The foregoing examples illustrate only a few embodiments of the application and are described in detail herein without thereby limiting the scope of the application. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the application, which are all within the scope of the application. Accordingly, the scope of the application should be assessed as that of the appended claims.
Claims (10)
1. A method of selecting an interpretable method of a partial discharge classification model, the method comprising:
obtaining partial discharge classification models under different neural network architectures, and explaining the partial discharge classification models by utilizing a plurality of interpretable methods to obtain interpretation information corresponding to each interpretable method, wherein the interpretation information is used for representing the contribution degree of each attribute characteristic in the partial discharge classification models to the partial discharge classification result;
acquiring a preset interpretability result corresponding to each partial discharge classification model, and comparing the interpretation information corresponding to the interpretability method with the preset interpretability results corresponding to the various partial discharge classification models for each interpretability method to acquire a comparison result;
a target one of the plurality of interpretive methods is selected based on the comparison result.
2. The method of claim 1, wherein the obtaining the preset interpretable results for each of the partial discharge classification models comprises:
carrying out feature extraction on partial discharge ultrahigh frequency data under different fault types to obtain initial feature data of various attribute features;
acquiring the association degree of each attribute feature and the fault type based on the initial feature data under the different fault types, and acquiring at least one target attribute feature in each attribute feature based on the association degree;
and for each target attribute feature, disturbing feature data corresponding to the target attribute feature in the initial feature data to obtain disturbance feature data corresponding to the target attribute feature, and determining the preset interpretability result corresponding to each partial discharge classification model by combining the initial feature data and the disturbance feature data.
3. The method of claim 2, wherein the determining the preset interpretable result for each partial discharge classification model in combination with the initial characteristic data and the perturbation characteristic data comprises:
determining a target contribution degree of the target attribute feature to the partial discharge classification model by combining the initial feature data, the disturbance feature data and the partial discharge classification model;
and aiming at each partial discharge classification model, sequencing each target contribution degree by each target attribute feature to the target contribution degree of the partial discharge classification model to obtain a target sequencing result of each target attribute feature in the partial discharge classification model, wherein the target sequencing result is used for representing the preset interpretability result corresponding to the partial discharge classification model.
4. A method according to claim 3, wherein said comparing said interpretation information corresponding to said interpretation method with preset interpretation results corresponding to the respective partial discharge classification models, to obtain comparison results, comprises:
comparing interpretation sub-information corresponding to the partial discharge classification model in the interpretation information with the preset interpretability result aiming at each partial discharge classification model to obtain a comparison sub-result, wherein the interpretation sub-information is used for representing the contribution degree of each attribute feature in one partial discharge classification model to the partial discharge classification result;
and taking the average value of the comparison sub-results corresponding to the partial discharge classification models in the interpretation information of the interpretability method to obtain the comparison result of the interpretability method.
5. The method of claim 4, wherein comparing the interpretation sub-information corresponding to the partial discharge classification model in the interpretation information with the preset interpretability result to obtain a comparison sub-result comprises:
obtaining the similarity degree of the interpretation sub-information and the preset interpretability result;
and obtaining a comparison sub-result according to the similarity.
6. The method of claim 5, wherein the obtaining the similarity between the interpretation sub-information and the predetermined interpretable result comprises:
judging the sequencing result of each target attribute feature in the interpretation sub-information according to the contribution degree of each attribute feature in the interpretation sub-information to the partial discharge classification result;
and obtaining the similarity according to the target sequencing result in the sequencing result and the preset interpretability result.
7. A selection device for an interpretable method of a partial discharge classification model, the device comprising:
the system comprises an acquisition module, a calculation module and a calculation module, wherein the acquisition module is used for acquiring partial discharge classification models under different neural network architectures, and utilizing a plurality of interpretable methods to interpret various partial discharge classification models to obtain interpretation information corresponding to each interpretable method, wherein the interpretation information is used for representing the contribution degree of each attribute characteristic in various partial discharge classification models to partial discharge classification results;
the comparison module is used for acquiring preset interpretive results corresponding to each partial discharge classification model, and comparing the interpretation information corresponding to the interpretive method with the preset interpretive results corresponding to the various partial discharge classification models for each interpretive method to acquire comparison results;
and the selection module is used for selecting a target interpretability method in the plurality of interpretability methods according to the comparison result.
8. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any of claims 1 to 6 when the computer program is executed.
9. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 6.
10. A computer program product comprising a computer program, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311024710.3A CN117034083A (en) | 2023-08-14 | 2023-08-14 | Method, device and equipment for selecting interpretation method of partial discharge classification model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311024710.3A CN117034083A (en) | 2023-08-14 | 2023-08-14 | Method, device and equipment for selecting interpretation method of partial discharge classification model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117034083A true CN117034083A (en) | 2023-11-10 |
Family
ID=88625985
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311024710.3A Pending CN117034083A (en) | 2023-08-14 | 2023-08-14 | Method, device and equipment for selecting interpretation method of partial discharge classification model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117034083A (en) |
-
2023
- 2023-08-14 CN CN202311024710.3A patent/CN117034083A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Kim et al. | Genetic algorithm to improve SVM based network intrusion detection system | |
| Shim et al. | Active cluster annotation for wafer map pattern classification in semiconductor manufacturing | |
| CN110022313B (en) | Polymorphic worm feature extraction and polymorphic worm identification method based on machine learning | |
| CN114266284A (en) | Method, device, equipment and program product for detecting insulation defect type of switch cabinet | |
| CN113946983A (en) | Method and device for evaluating weak links of product reliability and computer equipment | |
| CN116611003A (en) | Transformer fault diagnosis method, device and medium | |
| CN116595363A (en) | Prediction method, apparatus, device, storage medium, and computer program product | |
| CN116304713A (en) | Wind power plant fault scene prediction model generation method and device and computer equipment | |
| CN117034083A (en) | Method, device and equipment for selecting interpretation method of partial discharge classification model | |
| CN116191398A (en) | Load prediction method, load prediction device, computer equipment and storage medium | |
| CN114239639A (en) | Feature selection method and device, computer equipment and storage medium | |
| CN114417977A (en) | Fault diagnosis method and system based on transformer data sample expansion | |
| CN114580761A (en) | Steam quantity abnormity determining method and device, electronic equipment and readable storage medium | |
| CN115829543B (en) | Method for determining validity of preventive test of power equipment based on fault detection interval | |
| CN117192373B (en) | Power battery fault analysis method, device, computer equipment and storage medium | |
| CN115913699B (en) | Power distribution network lateral access detection method, device, computer equipment and storage medium | |
| CN117349098A (en) | Graphics processor performance test method, device, computer equipment and storage medium | |
| CN116884442A (en) | Emotion recognition method, emotion recognition device, computer equipment and storage medium | |
| CN117278381A (en) | Processing method, device, computer equipment and medium for server alarm data | |
| CN118134099A (en) | Equipment monitoring method, device, equipment, storage medium and computer program product | |
| CN118277271A (en) | Abnormality locating method, abnormality locating device, computer device and storage medium | |
| CN117875233A (en) | Method and device for determining chip power consumption related parameters and computer equipment | |
| CN118228154A (en) | Power grid frequency anomaly detection method, device, computer equipment and storage medium | |
| CN117135663A (en) | Abnormality identification method and device for base station energy saving index data, computer equipment and storage medium | |
| CN117290196A (en) | System fault prediction method, model training method, device and computer equipment |
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 |