Research On Power Transformer Fault Diagnosis System Based On Case-based Reasoning

As one of the most important equipment in power system, the running state of powertransformer directly affects the economic operation and safe production of power grid. Inorder to guarantee the safe and reliable operation of the power transformer, the research onfault diagnosis about it is very necessary.This paper establishes the transformer fault diagnosis model based on Case-BasedReasoning(CBR). Firstly, Through analyzing transformer fault principle, fault characteristic,fault effect and corresponding measures to be taken and integrating a large number of faultcase information, the fault case base was established using object oriented case representationmethod. Next, the corresponding similarity calculation model was designed aiming at the faultdata of different types. The special data was processed to fit the case reasoning rules ofoperation. Then, for the core of CBR, case retrieval, this paper proposes the improved K-NNalgorithm based on entropy and clustering. It evaluates the classification ability and theimportant degree of attributes using information gain which is used as the basis for the choiceof properties. Introducing the idea of clustering, this improved KNN algorithm extracts thethe most relevant attribute weights about the problem. And then, this paper analyzes therelationship between gas chromatography analysis and internal fault of transformer, aiming atthe complex dispersion characteristics of transformer fault DGA data, support vectormachine(SVM) with kernel function was designed to create the transformer fault diagnosismodel. At the same time, the Fruit Fly Optimization Algorithm(FOA) was used to optimizethe key parameters of SVM model and the transformer fault diagnosis model of FOA-SVMwas created. Compared with the heuristic algorithm to optimize the parameters of SVM,FOA-SVM is better at the fault classification performance. Finally, a system for faultdiagnosis model based on CBR was built to diagnose the power transformer.In the end, the transformer diagnosis system based on CBR was built to simulate inVisual C++6.0, Matlab2010a programming environment, using the database technology.Results show us the idea designed in this paper reasonable and effective.