GA-SVM-BP-based voltage transformer fault diagnosis method

Abstract

The invention discloses a GA-SVM-BP-based voltage transformer fault diagnosis method. Each type in a collected sample set S= {(x1, y1), (x2, y2), ... (xn, yn)} having type labels for an oil-immersed voltage transformer each class is divided into training samples and test samples by a 3:1 ratio; xi represents five sample attributes comprising hydrogen, methane, ethane, ethene and acetylene while yirepresents type labels 1, 2, 3, 4, 5 and 6 that respectively correspond to a normal state, middle temperature overheat, high temperature overheat, partial discharge, spark discharge and arc discharge; a DAG-SVM voltage transformer fault diagnosis model and a BP neural network are built first, a GA-DAG-SVM model and a GA-BP neural network are then established, the GA-DAG-SVM model is combined withthe GA-BP neural network, and fault diagnosis can be conducted on the voltage transformer. Faults of the voltage transformer can be accurately diagnosed via the method disclosed in the invention.

Description

technical field [0001] The invention belongs to the technical field of transformer fault online monitoring methods, and in particular relates to a transformer fault diagnosis method based on GA-SVM-BP. Background technique [0002] In recent years, the smart grid is the goal and direction of the development of my country's power sector and grid companies, and intelligent fault diagnosis of the operating status of transformers is also an inevitable trend. [0003] The fault diagnosis method of oil-immersed transformer has developed from the initial regular maintenance to the current online monitoring, which is mainly based on the dissolved gas in oil (DGA) combined with the existing intelligent algorithm for transformer fault diagnosis. At present, there are many intelligent algorithms applied here, such as: the relatively mature BP neural network, the combination of classic bagging and other algorithms, decision tree, etc. The application of these algorithms has greatly prom...

AI Technical Summary

Problems solved by technology

At present, there are many intelligent algorithms applied here, such as: the relatively mature BP neural network, the combination of classic bagging and other algorithms, decision tree, etc. The application of these algorithms has greatly promoted the research of transformer fault diagnosis. function, but it also has its own shortcomings, such as: BP neural network has high requirements for sample quality, Bagging algorithm may produce unrecognizable results, and decision tree is prone to overfitting problems

Images