Intelligent Fault Diagnosis And Pattern Recognition Of Oil-Immersed Transformer

Power transformer is the most important power transmission and transformation equipment. It is also the key equipment to ensure the trouble-free operation of the power system. How to ensure the safe and stable operation of the transformer and how to improve the reliability of power supply are important issues for the electric power enterprises. Dissolved Gases Analysis(DGA) is an important means to find latent fault of oil immersed equipment,and it is also an important basis to diagnose the internal fault of transformer. Due to the limitations of the traditional diagnosis method and the fuzzy and complex relationship between fault symptoms and fault mechanism, which led to the diagnosis accuracy is not high. With the rapid development of computer technology and artificial intelligence theory, a new approach of transformer fault diagnosis has opened up. This paper uses three kinds of intelligent methods to diagnose transformer fault. Three intelligent methods are as follows:(1) Based on the interval arithmetic of IEC recommended three ratio method, including fuzzy diagnosis and matter-element extension diagnosis. The focus of the study is membership function and dependent function;(2) To carry out self learning neural network diagnosis, using data field potential function improved convergence of BP neural network and construct diagnostic topology of the neural network;(3) Through the establishment of transformer fault oil-gas database, to study stay diagnosis data and sample data in the database similarity between degree, using fuzzy cluster to diagnose. A set of transformer intelligent monitoring is developed by using the MFC application software.This system can summarize a variety of intelligent diagnosis results and obtain the diagnosis results of the nature and components of the transformer fault by combining fuzzy comprehensive assessment method.Finally, the accuracy of the results is verified by examples.