Research On Fault Diagnosis Of Rolling Bearing Based On Acoustic Signal Morphological Component Analysis

Vibration can lead to mechanical noise, this is to say mechanical acoustic signal is the extension of vibration signal. And acoustic property varies with mechanical vibration. Acoustic fault diagnosis belong to the non-destructive testing field, it has many characteristics such as equipment operation without disturbing, simple operation and so on. In traditional acoustic fault diagnosis, the impact fault of acoustic signal resulted from bearing early damage is easy to be covered up by different noises.Therefore, it is difficult to discriminate the bearing fault type and guarantee the accuracy and reliability of fault diagnosis results. Aimed to solve the problem in this paper, a new method of rolling bearing's fault diagnosis based on MCA and Hilbert spectrum analysis of acoustic signal is given with mastering the basic acoustic knowledge, characteristics of rolling bearing's acoustic fault signal and sparse characteristics of redundant dictionary.On one hand, not only the fault features of race defect but also the acoustic theory are analyzed.The effective sound pressure fields of the holographic plane and their projection are used as a failure criterion of mechanical equipment. On the other hand, by analyzing the features of wavelet function, a redundant dictionary consists of coif4 dictionary and local cosine dictionary which can achieve the necessary flexibility is built for original signal's sparse representation according to the frequency characteristics of acoustic fault signal from single-channel microphone. In addition, Hilbert spectrum is carried out to analyze the impact component after sparse decomposing the original acoustic signal by MCA.After that, a simulation, including cross-correlation analysis of impact component before and after signal simulation processing, is conducted to verify the new method's validity on bearing fault type recognition mathematically. Besides, a rolling bearing fault diagnosis system with microphone array is built and some experiments utilizing different microphone arrays are conducted with three kinds of bearings, including fault-free, inner race fault and outer race fault bearings, at semi-anechoic room and ordinary lab, relatively. The hot spots in holographic plane3 D effective sound pressure fields and their projection measured by different microphone arrays are used operation state evaluation of the test-bed. The time-frequency analysis of hot spots corresponding to single-channel microphone's signal from key position of outer race fault bearings is done, and the accuracy and reliability of bearing fault diagnosis by acoustic signal is further proved. What's more,the new method is used to process signal of inner ring fault and its fault type is recognized accurately based on the analysis of time-domain ?time-frequency plane and Hilbert spectrum of the impact component. The results of simulations and experiments proved that this method can identify the types of rolling bearings' faults accurately. Hence, this study provides a new method for fault diagnosis technology in acoustics.