| The electric drive system of Hybrid Electric Vehicle (HEV) has a complex work environment. HEV electric drive system worked on the urban condition is required to be start and stop frequently, which cause a great fluctuation to the electric drive system.So HEV electric drive system breakdown is easily happened. In order to solve the problem, fault diagnosis methods of HEV electric drive system are researched.Wavelet Packet Theory is used to analysis the multiple signal data which are collected from the HEV. By Wavelet Packet Decomposition, all kinds of potential fault information of HEV electric drive system can be discovered, especially those faults that hasn’t shown or happened at early stages. It is helpful to find the failures and make up in time, which guarantees vehicle and traffic safety. The Kernel Principal Component Analysis(KPCA) is adopted to simplify the fault characteristics of HEV electric drive system. The characteristic matrix of HEV electric drive system is reduced and the row information is restored. The Nuclear methods and the Principal Component Analysis method is combined to guarantee the safety of non-linear space information. During the fault classification, the method Fuzzy Kernel Clustering analysis is proposed, which has the advantages of accuracy and celerity.The research of HEV electric drive system fault diagnosis methods based on Wavelet packet decomposition, the Kernel Principal Component analysis, and Fuzzy Kernel Clustering makes the process of fault diagnosis rapid and accurate. |
