Abnormity detection and evaluation method for main transformer synthetic data set

By calculating the mutual information values ​​between the input variables and the target variables, and optimizing the DBN structure with GB-RBM and GA genetic algorithms, using the K-means clustering method to identify abnormal data, solving the problem of inaccurate capture of complex abnormal patterns in the existing technology, and achieving more efficient and accurate abnormal detection.

CN119720014APending Publication Date: 2025-03-28GUANGDONG POWER GRID CO LTD DONGGUAN POWER SUPPLY BUREAU
0 Cites 0 Cited by

Patent Information

Application Number
CN202411768164.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-12-04
Publication Date
2025-03-28

AI Technical Summary

Technical Problem

The existing main transformer abnormality detection and evaluation methods rely on simple feature extraction methods and models that cannot effectively adapt to the actual data distribution, resulting in the inability to accurately capture complex abnormal patterns in the data, affecting detection accuracy, and overfitting may occur, resulting in wasted computing resources.

Method used

By calculating the mutual information values ​​between the input variables and the target variables, filter out input variables with high correlation with the label variables, build an input matrix, and use the GB-RBM algorithm for feature extraction. Combined with the GA genetic algorithm, the number of hidden layer nodes of DBN is optimized, and the abnormal data is identified and separated by K-means clustering method, and the effect of abnormal detection is finally evaluated through the clustering density and resolution.

Benefits of technology

It improves the sensitivity of the model to abnormal patterns, enhances the accuracy of detection, reduces the risk of overfitting, improves the adaptability and efficiency of the model, and ensures the reliability and effectiveness of the detection results.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention discloses an anomaly detection and evaluation method for a synthetic data set of a main transformer, and relates to the technical field of transformer state detection, and the method comprises the following steps: obtaining monitoring data and a diagnosis result of a transformer substation, combining the monitoring data and the diagnosis result into two data sets, one is used for feature input, and the other is used for a target variable; calculating mutual information values between the input variables and the target variables, and based on the calculated mutual information values, screening out the input variables having relatively high correlation with the tag variables for constructing an input matrix; organizing the screened input variables into an input matrix, performing feature extraction on the input matrix by adopting a GB-RBM algorithm, and then optimizing the number of hidden layer nodes of the DBN by applying a GA genetic algorithm; performing K-means clustering on the feature vectors extracted by the DBN, and identifying and separating abnormal data; analyzing a clustering result, and marking abnormal data; parameters in the feature extraction and clustering process are adjusted according to an evaluation result, and the accuracy and efficiency of anomaly detection are further improved.
Need to check novelty before this filing date? Find Prior Art