Research On Transformer Fault Diagnosis Based On New Fuzzy Clustering Algorithm

Power transformer is one of the pivot equipment of power system, Its normal operation is the important guarantee for the safe operation of power system, Research on the fault diagnosis of power transformer is of important theoretical value and realistic significance. In the fault diagnosis of transformer, the relationship between the fault cause and fault symptom is fuzzy, Fuzzy c-means clustering algorithm can classify fault samples in a fuzzy way, on that basis, this paper proposes new algorithms is proposed to improve the accuracy of fault diagnosis of transformer.In the insulation fault of transformer, different types of fault produce different main characteristic gasses and the components. The samples which are composed of the components of dissolved gas-in-oil are present in variable degrees, which should be treated differently. This paper presents a weighted fuzzy c-means clustering algorithm which utilizes the weights to express the relative degree of the importance of various samples in fault classification. Compared the diagnostic findings with Fuzzy c-means clustering algorithm, the proposed algorithm has obviously improved the accuracy and robustness of fault classification.On the basis of the above, this paper investigates a dynamic weighted fuzzy c-means clustering algorithm based on genetic algorithm. The algorithm adopts a kind of cluster-center-based floating point encoding mode, In which the variable length chromosomes express cluster prototypes and different length of chromosomes corresponding to different numbers of cluster prototypes, The algorithm iterates in a new type of crossover operator and mutation operator. The experimental results show that the algorithm not only overcomes the shortcoming which fuzzy c-means clustering algorithm is the sensibility to initial value, but also can scientifically reflect the real structure of fault sample data, Therefore, The algorithm improves the accuracy of fault classification.