Research On Fault Prediction Method Of Power Electronic Circuit

With the development and increasing requirements of security and reliability of aviation and aerospace industry, much more attention is being paid to aircraft health management technology as an effective way to improve the security. The fault prediction of aircraft power system is an important part of aircraft health management and the key of it lies in the fault prediction of electronic circuits, thus the research of fault prediction methods on power electronic circuits is of great significance.This paper gives an overview of the significance, research status at home and abroad, the difficulties and trend of the fault prediction technology on power electronic circuits. The failure rate and fault modes of power electronic devices are introduced and the main characteristic of electrolytic capacitor which has the highest failure rate is analyzed. The characteristic parameter extraction methods of electrolytic capacitor are studied, which are time-domain analysis method, frequency domain analysis method based on FFT and ripple voltage method. Then the methods are used to extract the characteristic parameter of electrolytic capacitors working in Buck and Boost converter respectively, the results of the experiments show the validity. For the purpose of achieving the parameter identification of power electronic circuits, the methods based on hybrid systems theory, nonlinear least squares theory and genetic algorithm theory are studied. To prove the validity of the three methods, experiments of Buck converter are designed. The extraction methods of system-level characteristic parameter of power electronic circuits are also researched, and verified and evaluated on Buck converter. On the basis of all the work done before, by employing the prediction methods of LS-SVM, BP neural network and AR model, the fault prediction of the electrolytic capacitor and Buck converter is achieved. At the end of this paper, the research is summarized and its further direction is pointed out.The work presented in this paper is funded by National Natural Science Foundation of China (60871009) and Aeronautical Science Foundation of China (2009ZD52045).