Fault Diagnosis Of Rotating Machinery Based On HHT And Netural Network

Rotating machine is the key equipment of the aviation, power, chemical and many other aeras, sothe fault diagnosis has an important practical significance. With the continuous evolution of vibrationdetection, signal processing and other related technology, it has became an important researchdirection of fault diagnosis based on the vibration signal detection, processing and analyze.Meanwhile, the development of neural network, genetic algorithms and many other theories has madea new way to research and apply of fault diagnosis.This paper describes Hilbert-Huang transform (referred to as HHT) method, neural network andother related contant. On the one hand, it describes the basic principle and implementation process ofHHT method, and analysis the existing endpoint effect and the false mode problem; on the other hand,it describes the basic theory of the BP neural network and genetic algorithm, and researches theprocess of genetic algorithm BP network optimization as well, then the BP network is optimized forits insufficient by using genetic algorithm.Meanwhile, using multi-rotor test stand simulates common rotating machine fault, and usingHHT method processes and analyze the failure for studing the rotating machine fault diagnosis, thenas the fact that Fuzzy Entropy can express complexity of signal and have relative stability, the FuzzyEntropy theory is introduced to fault diagnosis, then a method of feature extraction is proposed basedon a combination of EMD and fuzzy entory, and is taken to fault diagnosis of rotor system, which isfeasible and effective.Finally, using HHT method and BP neural network optimized by genetic algorithm to researchfault diagnosis of rotor system integratedly. The common fault characteristics for rotor system isextracted, and is put into BP network model optimized by the genetic algorithm for fault diagnosis.The result indicates that this method can diagnosis the fault of rotor system effectively.