Oil-filled electrical equipment fault diagnosis method based on gas relation and graph neural network

Abstract

The invention relates to the field of artificial intelligence, and discloses an oil-filled electrical equipment fault diagnosis method based on a gas relationship and a graph neural network, wherein the method comprises the steps: step 1, constructing a graph neural network model based on oil-filled electrical equipment fault diagnosis, and building a characteristic gas relationship model; step 2, updating the edges of a graph by using the information of the topological structure of the characteristic gas relation graph and the information of characteristic gas nodes; step 3, updating gas nodes in the graph by utilizing topological structure information of the characteristic gas relation graph and the updated edge information; and step 4, establishing a linear updating and classification layer, and outputting a fault type. The graph neural network model is constructed by using the correlation graph of the dissolved gas in the oil of the oil-filled electrical equipment and the artificial intelligence technology, fault diagnosis of the oil-filled electrical equipment is realized, and compared with a traditional fault diagnosis method of oil-filled electrical equipment, the fault diagnosis method has higher accuracy in the aspect of fault diagnosis, and has generalization for diagnosis and discrimination of various faults.

Description

Technical field [0001] The present invention relates to a fuel-filled electrical equipment fault diagnosis method, and more particularly to a filling electrical equipment fault diagnosis method based on a gas relationship and a nerve network. Background technique [0002] The fuel-filled electrical equipment fault diagnosis method described in the existing literature is mainly divided into two links. One is to extract abnormal characteristics generated by the fuel-fuel electrical equipment failure, such as feature gases, infrared rays and temperatures, etc., according to these abnormal features. Summary analysis results in the relationship between the corresponding fault characteristics. At present, the diagnosis of fuel-filled electrical equipment is mainly based on the abnormal characteristics of the characteristic gas content dissolved in oil. As the fuel-filled electrical equipment has long run, the internal insulation gradually loses, and the hydrocarbon compound in the insu...

AI Technical Summary

Problems solved by technology

In the method of fault diagnosis of oil-filled electrical equipment based on characteristic gas concentration using machine learning, for example, BP neural network and support vector machine only reflect the mapping relationship between the characteristic gas concentration and the fault type of oil-filled electrical equipment, without taking into account The connection between the concentration of each characteristic gas has the phenomenon of insufficient mining and extraction of the content characteristics of the characteristic gas of oil-filled electrical equipment, which may lead to inaccurate failure analysis of oil-filled electrical equipment, resulting in low accuracy of fault diagnosis of oil-filled electrical equipment

Images