Improved Fault Diagnosis Of Rotating Machinery Based On WVD

Time-frequency analysis method is an important branch of non-stationary signal processing, which makes use of the joint time and frequency function to represent the non-stationary signals, and carries on the analysis and processing. The time-frequency analysis method according to the different time-frequency joint function, can be divided into linear time-frequency representation and bilinear time-frequency representation. Bilinear time-frequency representation is also called quadratic time-frequency representation, it reflects the time-frequency distribution of signal energy, Winger-Ville distribution is one of the most famous, it has better time-frequency resolution, and can better express the time-varying characteristics of the signal, and has a broad application prospect in rotating machinery fault diagnosis.As the WVD belongs to the quadratic time-frequency representation of signals, it would inevitably produce cross interference, seriously influences the practical result. This paper in view of the WVD cross interference problems has a extensively studied, firstly it summarizes the existing inhibition technology, and then make it into three categories which are energy distribution method, signal decomposition method and superposition of spectra method according to the principle of the inhibition. Then this paper proposes a WVD method based on EMD-ICA, and makes a lot of simulation experiment. The experimental results showed that the method keeps the WVD higher resolution at the same time, effectively inhibit the emergence of cross interference item. Experimental results show the effectiveness of the proposed method, and transverse comparison with other methods of suppression, proved the superiority of the algorithm.This paper uses the improved method of WVD method to diagnosis the rotating machinery fault characteristic information in order to categorize the type of fault and monitoring. In the end this paper analyzed the four kinds of signals collected from a reduction gear box bearing. The improve WVD method can clearly distinguish the fault frequency characteristics, so as to determine the fault location and accurate to reflect the operation state of the machine.This article research results show that the WVD method based on EMD-ICA improved effectively for rotating machinery fault signal diagnosis, it make an accurate evaluation about its running condition, if fails, it can judge the fault type, the method is worthy of popularization and application.