Study on Fault Feature Extraction of High-Speed Automaton

In order to effectively extract the weak fault feature of high-speed automaton (HSA) in the environment with strong noises, a method of fault feature extraction was proposed based on wavelet packet energy entropy and fuzzy clustering algorithm. In this paper, wavelet packet was utilized to denoise the vibration signals of three working conditions of automaton, to decompose the signals and then to obtain eight frequency band energy entropies of each signal. Through processing and analyzing the features, the results show that there are obvious differences between three conditions, and the fuzzy clustering algorithm can identify the fault pattern of HSA accurately. The feature proposed by this extraction approach is proved to be able to effectively reflect the working state of the automaton, therefore the wavelet packet energy entropy could be considered as the feature parameters of HAS for fault identification and diagnosis. The fault feature extraction method can also provide a certain engineering application value for real-time monitoring and early fault diagnosis of this type HSA.