High-speed Automaton Fault Diagnosis based on Wavelet Permutation Entropy

Abstract

This paper studies the automaton vibration response signal with wavelet packet transform for time-frequency analysis, put forward to each layer of wavelet packet coefficients doing wavelet scale spectrum rearrangement process, each layer of wavelet coefficients are calculated by the permutation entropy (PE), as automaton when short transient impact the weak fault signal characteristics. Finally using support vector machine (SVM) to fault classification recognition characteristics, the results show that this method can effectively extract the characteristic value and identify the weak fault, can effectively solve the problem of automaton fault diagnosis.