Running Status Diagnosis of Onboard Traction Transformers Based on Kernel Principal Component Analysis and Fuzzy Clustering

The onboard traction transformer is a critical equipment of high-speed trains, its running state directly affects the safety and stability of a train's operation. Given the complexity of the running condition of the onboard traction transformer, this paper proposes a running state diagnosis algorithm based on kernel principal component analysis (KPCA) and fuzzy clustering. To fully extract the status information of the onboard traction transformer, the aging characteristics of insulating oil and main insulation are analyzed under different running mileage as the first step. Thereby, to eliminate the signal redundancy, the status feature set of the onboard traction transformer is analyzed by KPCA combined with the characteristic quantities of the traditional dissolved gas analysis (DGA), and the eigenvalues with the contribution rate of over 95% are used as new eigenvectors. Finally, a status diagnosis model is established by using fuzzy clustering analysis, considering the limitations of fault data of onboard traction transformer. The results from field collected data show that the proposed method is effective in diagnosing the running status of the onboard traction transformer.