Fault Diagnosis of Rolling Bearing based on Permutation Entropy Optimized Maximum Correlation Kurtosis Deconvolution

Abstract

An improved MCKD algorithm was proposed as an adaptive parameter selection approach. In this method, the optimal filter length of the MCKD algorithm was determined by the maximum permutation entropy value and the optimal fault period of the MCKD algorithm was determined by the maximum kurtosis value. The determined optimal filter length and optimal fault period were verified by simulation signals of the bearing as well as fault experimental signals of the inner and outer rings. Results demonstrate that in early fault diagnosis, the improved MCKD algorithm based on permutation entropy can extract faint characteristic information of early fault of rolling bearing effectively. It can highlight fault pulse signals which are covered by noises and get more ideal results than the minimum entropy deconvolution algorithm.