Spatial Time Frequency Distribution Method And Its Application In Mechanical Fault Diagnosis

In the actual industrial environment,the gearbox has high working intensity and bad working environment,which is easy to failure in the process of use.Analyzing the vibration signal of gearbox is a practical and effective way to detect faults.However,the vibration signals collected by sensors contain not only the coupling vibration information of key parts such as gears and bearings,but also background noise and other interference information.Especially when a variety of gearbox faults occur at the same time,it is more difficult to detect and diagnose the gearbox faults.In order to solve the problem of multi-fault in mechanical equipment,an underdetermined blind source separation method based on space time frequency distribution(STFD)is proposed and applied to multi-fault diagnosis of bearings.The specific research contents are as follows:(1)Firstly,the basic theory of blind source separation and two mathematical models of linear mixing and convolution mixing are studied;secondly,a classical blind source separation algorithm Fast ICA algorithm is analyzed.Then,the common fault types,causes and their characteristics in time-frequency domain are analyzed.(2)When blind source separation(BSS)algorithm is used to separate fault signals,it is usually necessary to accurately estimate the number of fault sources.In order to solve this problem,according to the characteristics of bearing fault signal in time-frequency domain,a bearing fault sources number estimation algorithm based on compact kernel distribution(CKD)and peak search algorithm is designed.Firstly,the time-frequency distribution of the signal is obtained by using the compact kernel distribution,and then the energy distribution of the signal in the frequency domain is obtained by summing the time-frequency distribution in the time domain.The frequency bands corresponding to the higher energy is found by peak search algorithm,and the frequency band is compared with the characteristic fault frequency of bearings to preliminarily determine the number of fault sources in the observed signal.Finally,the effectiveness of the method is verified by simulation experiments.(3)In view of the fact that the number of fault sources is larger than the number of sensors,an underdetermined blind source separation method based on STFD is proposed.Firstly,the STFD matrix of the observed signal is obtained by CKD.After preprocessing the STFD matrix,the time-frequency distribution of the fault sources is obtained by automatic item selection and clustering.Then,according to the inversion characteristics of Wigner Ville distribution(WVD),the time-frequency distribution is used to reconstruct the source signals.Finally,the simulation results show that the method has a good effect.(4)The experiments are designed and carried out,and the measured data are obtained.In this paper,the fault source number estimation method based on CKD and peak search algorithm and the underdetermined blind source separation method based on STFD are combined to analyze the measured signals.Experimental results show that the proposed method can effectively estimate the number of fault sources and separate fault signals.Compared with Fast ICA algorithm,the effect of the proposed algorithm is better.