CN-121980367-A - Power system fault classification method, device, equipment and program product

CN121980367ACN 121980367 ACN121980367 ACN 121980367ACN-121980367-A

Abstract

The application is applicable to the technical field of power systems and provides a power system fault classification method, a device, equipment and a program product, wherein the method comprises the steps of acquiring electrical quantity time sequence data acquired by power system monitoring equipment; the method comprises the steps of carrying out multi-level signal decomposition on the electrical quantity time sequence data to obtain a plurality of signal components, extracting statistical characteristics of the plurality of signal components to construct feature vectors to be classified, inputting the feature vectors to be classified into a pre-trained breadth feature mapping classification model, outputting a fault classification result, and effectively distinguishing complex fault types and improving classification accuracy and fineness.

Inventors

XU YULIN
CHEN YIFANG
YAO HAN
CAI QINGQING
YANG JIAN
JIANG DING
WANG ZHAOWEI
Rao Wenqiang
HUANG YIFENG
CHENG DEPING
YANG HAO
ZHU HUIWEN
She Hanmin
HU JINHU
YU WEIGUO
ZHANG XIAO
WU QUANWEI
CHEN TAO
Cao Houpu
Long Mengfei
LI FAN

Assignees

长园深瑞继保自动化有限公司

Dates

Publication Date: 20260505
Application Date: 20251203

Claims (10)

1. A method of power system fault classification, comprising: Acquiring electrical quantity time sequence data acquired by power system monitoring equipment; Carrying out multi-level signal decomposition on the electrical quantity time sequence data to obtain a plurality of signal components; extracting statistical features of the plurality of signal components to construct feature vectors to be classified; and inputting the feature vector to be classified into a pre-trained breadth feature mapping classification model, and outputting a fault classification result.
2. The method of claim 1, wherein said performing multi-level signal decomposition on said electrical quantity timing data further comprises: And performing data dimension reduction processing on the acquired electrical quantity time sequence data.
3. The method of claim 1, wherein said performing multi-level signal decomposition on said electrical quantity timing data to obtain a plurality of signal components comprises: Iteratively decomposing the electrical quantity time sequence data by using a decomposition kernel function, and generating a compression characteristic variable and a redundant information component in each layer of decomposition; And taking the compression characteristic variable of the upper layer as the input of the decomposition of the lower layer, and taking the compression characteristic variable obtained in the last iteration and the redundant information component generated in the previous iteration as the signal components after a plurality of iterations.
4. The method of claim 1, wherein the statistical features include at least two of energy, variance, kurtosis, and information entropy.
5. The method of claim 1, wherein the breadth feature mapping classification model maps the feature vectors to be classified into high-dimensional combined features through a breadth feature mapping layer and classifies based on the high-dimensional combined features by a decision layer.
6. The method of claim 5, wherein the training process of the breadth-feature-mapping classification model is: The sample feature vector marked with the fault type label is used as a training set to be input into the breadth feature mapping layer, so that feature mapping output Zr and breadth enhancement output Hr are obtained; Splicing the feature mapping output Zr and the breadth-enhancement output Hr to obtain a combined feature matrix Ar; inputting the combined feature matrix Ar into the decision layer to obtain decision output Dr; and solving an output weight Wo according to the decision output Dr and a fault type label Yr corresponding to the training set, and completing the training of the breadth feature mapping classification model.
7. The method of claim 6, wherein inputting the feature vector to be classified into a pre-trained breadth feature mapping classification model, outputting a fault classification result, comprises: Inputting the feature vector to be classified into a breadth feature mapping layer of the breadth feature mapping classification model to obtain a combined feature matrix Ae; And outputting the fault classification result according to the output weight Wo and the combined feature matrix Ae.
8. A power system fault classification device, comprising: the acquisition module is used for acquiring the electrical quantity time sequence data acquired by the power system monitoring equipment; the decomposition module is used for carrying out multi-level signal decomposition on the electrical quantity time sequence data to obtain a plurality of signal components; The feature extraction module is used for extracting statistical features of the plurality of signal components to construct feature vectors to be classified; The fault classification module is used for inputting the feature vector to be classified into a pre-trained breadth feature mapping classification model and outputting a fault classification result.
9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the method according to any of claims 1 to 7 when executing the computer program.
10. A computer program product, characterized in that the computer program product, when run on a computer device, causes the computer device to perform the method of any of claims 1 to 7.

Description

Power system fault classification method, device, equipment and program product Technical Field The application belongs to the technical field of power systems, and particularly relates to a power system fault classification method, device, equipment and program product. Background In a power system, fault diagnosis is a key link for guaranteeing safe and stable operation of a power grid. The basis for realizing accurate fault diagnosis is based on the acquisition and analysis of high-precision and high-frequency voltage, current and other electrical quantity waveform data when the power grid faults occur. Some existing fault diagnosis methods generally include acquiring electrical data, decomposing a signal into a plurality of components by a signal processing method, extracting characteristic parameters for each component, and finally performing fault identification by using a classifier. However, most of these methods rely on setting a fixed threshold value for a single index such as a current abrupt change, a voltage drop depth, etc., or rely on manually set logic rules to perform judgment, when the classifier structure is relatively simple or the rule setting is not perfect enough, when the classifier structure faces complex working conditions caused by new energy grid connection, increase of power electronic equipment, etc., especially for high-resistance ground faults or when signals are interfered by strong noise, the fixed threshold value is difficult to set, and erroneous judgment or missed judgment is easy to generate. In addition, these methods generally can only perform simple fault judgment, have poor classification capability for specific fault types, have weak generalization capability, and are difficult to adapt to changeable power grid operation modes. Therefore, how to diagnose and classify the fault state of the power grid rapidly, accurately and automatically is a technical problem to be solved urgently by those skilled in the art. Disclosure of Invention The embodiment of the application provides a power system fault classification method, a device, equipment and a program product, which aim to solve the technical problems of low fault classification accuracy, weak generalization capability and easy misjudgment and omission of judgment when facing complex working conditions due to dependence on fixed threshold values and manual rules in the prior art. In a first aspect, an embodiment of the present application provides a method for classifying faults of an electric power system, including: Acquiring electrical quantity time sequence data acquired by power system monitoring equipment; Carrying out multi-level signal decomposition on the electrical quantity time sequence data to obtain a plurality of signal components; extracting statistical features of the plurality of signal components to construct feature vectors to be classified; and inputting the feature vector to be classified into a pre-trained breadth feature mapping classification model, and outputting a fault classification result. In a possible implementation manner of the first aspect, the performing multi-level signal decomposition on the electrical quantity time sequence data further includes: And performing data dimension reduction processing on the acquired electrical quantity time sequence data. In a possible implementation manner of the first aspect, the performing multi-level signal decomposition on the electrical quantity time sequence data to obtain a plurality of signal components includes: Iteratively decomposing the electrical quantity time sequence data by using a decomposition kernel function, and generating a compression characteristic variable and a redundant information component in each layer of decomposition; And taking the compression characteristic variable of the upper layer as the input of the decomposition of the lower layer, and taking the compression characteristic variable obtained in the last iteration and the redundant information component generated in the previous iteration as the signal components after a plurality of iterations. In a possible implementation manner of the first aspect, the statistical features include at least two of energy, variance, kurtosis and information entropy. In a possible implementation manner of the first aspect, the breadth feature mapping classification model maps the feature vector to be classified into a high-dimensional combined feature through a breadth feature mapping layer, and classifies the feature vector based on the high-dimensional combined feature by a decision layer. In a possible implementation manner of the first aspect, the training process of the breadth feature mapping classification model is: The sample feature vector marked with the fault type label is used as a training set to be input into the breadth feature mapping layer, so that feature mapping output Zr and breadth enhancement output Hr are obtained; Splicing the feature mapping output Zr