Research On Rolling Bearing Fault Diagnosis Based On Enhanced Weak Fault Information

In most industrial machinery,rolling bearings are a crucial component.Once a bearing failure occurs,not only does it affect the normal operation of the entire machinery system,but it can also lead to serious consequences.Therefore,it is important to accurately monitor and diagnose the working condition of rolling bearings.The core problem in diagnosing bearing failures is to extract the characteristics of the failures from vibration data.However,vibration signals collected during complex machine operations are often filled with a lot of noise,and the failure characteristic information is masked.Traditional diagnostic methods usually cannot effectively extract these failure signals,which poses a great risk to the safety of equipment operation,such as bearing overheating,which can cause changes in the internal structure.Therefore,how to effectively diagnose rolling bearings under weak fault feature conditions is an important topic currently being studied by many mechanical fault diagnosis scholars.This paper will focus on rolling bearings and study topics such as noise suppression,enhancement of time-frequency resolution,and improvement of transient impact recognition rate.The main research contents are as follows:(1)A total variation robust local mean decomposition method is proposed to address the problem of strong background noise affecting the collected vibration signals of rolling bearings,as well as the low noise immunity and modal aliasing problems of the local mean decomposition method.The mathematical definition and theoretical derivation are given by combining the total variation denoising method with the robust local mean decomposition method.The total variation method is used to obtain a signal with a higher signal-to-noise ratio,and then the signal is subjected to local mean decomposition.The key parameters are selected using the fixed subset method,and the screening stopping criteria are set to adaptively determine the stopping conditions.Simulation and experimental results show that the proposed algorithm can effectively suppress noise interference,enhance weak fault information,and improve the system’s noise resistance ability.(2)An energy operator synchroextracting transform algorithm is proposed to address the problems of time-frequency ambiguity and transient feature error in the synchroextracting transform algorithm for complex time-varying signals.Firstly,the energy operator is considered to measure the instantaneous energy change of a signal composed of a single time-varying frequency,and is used as a pre-processing method,combined with the synchroextracting operator to extract the instantaneous fault feature frequency,and the corresponding method is theoretically derived.Then,Rényi entropy is selected as the performance index to indicate the concentration of time-frequency energy.Finally,the proposed method is analyzed and verified through rolling bearing fault experiments and simulation signals.The experimental results show that the proposed method can effectively enhance the time-frequency resolution,and has certain advantages compared with previous time-frequency analysis methods.(3)Aiming at the problem that the weak fault information of rolling bearings is often masked by pulse noise when analyzing transient impact,a successive variational mode decomposition method based on squared envelope spectrum is proposed.Firstly,successive variational mode decomposition is derived based on variational mode decomposition to reduce mode mixing phenomenon and computational complexity.Then,the true fault modal components are selected by using the weighted value of kurtosis index and cross-correlation coefficient KC.Finally,feature extraction is performed using the squared envelope spectrum.Experimental and simulation analyses show that the proposed method accurately identifies periodic transient impacts,effectively extracts weak features,and improves the accuracy and efficiency of fault diagnosis analysis for complex machinery.