Studies On On-line Supervision System Of Dissolved Gas In Transformer And The Fault Diagnosis Methods Based On Dissolved Gas Analysis

The study of fault diagnosis technology based on dissolved gas analysis(DGA) is very important to maintain the reliable running of electric power system, and also an efficient method to detect the incipient fault in transformer. Aiming at on-line detection of the oil in transformer, this paper mainly studies the design of control system about on-line detection system and the application of the complex neural network on the DGA fault diagnosis.The existed control systems about on-line detection of the dissolved gas in transformer all have many shortcomings, such as infirm expansibility, weak automatism and so on. To the questions, a new type of control system based on Can_bus was put forward, which hardware and software are also discussed how to realize.The popular fault diagnosis methods at present are all not enough accuracy and can't deal with complex fault type. In order to overcome the shortcomings of the popular methods, a new type of complex error back propagation neural network based on DGA is adopted, which syncretizes the tradition BP and RBF neural network and fully make use of the good ability of no-linear mapping character of neural network. At the same time,it simplifies the network and improves the contain error ability by reducing the propagation error and improving the verticality of input vector. Furthermore, a new data pretreatment method and an improved K-means arithmetic make the new neural network model more efficiency.Based on the complex error back propagation neural network model, this article has also developed a set of transformers fault diagnosis system with Visual Basic 6.0 as kit. The integrating experiment results of system testing-run proving the reliable of the control system and the validity of the new model.