Fault Diagnosis Of Lixing110kV Transformer Based On SVM

These years, the power system has got a tremendous progress with the development of smart grid. And the power transformers become more and more important with the scale-up of the grid capacity. Once the power transformers malfunction, it will have a negative impact on the industry and the development of national economy which threads the safety of people. Then the diagnosis of the malfunctions and the strategy of the overhaul in the power transformers have a crucial effect on the high quality of the power and the reliability of power supply.The object of the research in this thesis is the110kV main power transformers which is the brand of Lixing in the transformer substation. The insulation deterioration is the most common problem of the power transformers. With the development of the modern power electronics technology and the online testing, the test of insulating oil age; the analysis of oil dissolved gas chromatography; the test of partial discharge and the test of the transformer withstand voltage performance have become the common diagnosis of the malfunction in power transformer.In this thesis, we design the hardware unit of the malfunction monitoring system in power transformer based on the structure of the power transformer and mechanism of the malfunction. The TMS320F2812DSP is chosen as the main processor and we design other hardware circuit which includes the power supply module, processor interface, memory interface and the communication module. The monitoring system can meet the demand of the Lixing110kV main power transformer which needs the online test be real-time, flexible and fast.The diagnosis model of the malfunction in power transformer which is based on the support vector classifier machine uses the particle swarm optimization support vector machine. We use the toolbox of libSVM on Matlab to train the support vector classifier machine and use the machine to predict the fault conditions of the Lixing110kV main power transformer.The actual examples can prove that the result of diagnosis is the same as the fact. What’s more, this model is very simple and useful which means it can be promoted in the project and it can supply a reliable diagnosis technique in the Lixing110kV main power transformer.