Research On The Diagnosis Method Of Rolling Bearing Minor Faults Under Variable Speed Condition

As a key component of the nuclear main pump,the health status of rolling bearings directly affects the safety of nuclear power equipment,so it is of great engineering significance to diagnose their faults as early as possible.However,rolling bearing faults in the early stage often have the characteristics of low pulse amplitude and weak fault,which makes the fault diagnosis more difficult to a certain extent.In addition,in most cases,rolling bearings often operate under variable speed conditions,which leads to the inclusion of many complex interference components in the collected bearing signals,further increasing the difficulty of diagnosing microfaults in rolling bearings.Therefore,in response to the difficulty of diagnosing microfaults in rolling bearings under variable speed conditions,this paper takes rolling bearings under variable speed as the research object,and conducts research on methods of denoising,feature extraction,and fault recognition for bearing fault signals in sequence.And simulation and experimental analysis are conducted to verify the effectiveness of proposed methods.The specific research contents are as follows:In response to the problem of difficulty in denoising microfault signals of rolling bearings under strong noise and variable speeds,this paper proposes a denoising method based on the hunter prey optimization algorithm,variational mode decomposition and adaptive time varying morphological filter.Firstly,the HPO algorithm is proposed to optimize parameters of VMD.Secondly,the optimal components are selected by using the comprehensive index to reconstruct the signal.Then,the joint denoising work is completed by combining ATVMF.Finally,the envelope order spectrum is used to extract the bearing fault orders.And the theoretical order values of each bearing component are compared with the actual order values to achieve the microfault diagnosis under variable speed conditions.The effectiveness of the proposed method has been verified through simulation comparison and experimental analysis.To address the issue of traditional indicators being unable to extract features from variable speed bearing signals,this paper proposes a feature extraction method based on fractional hierarchical range entropy(Fr HRE).Firstly,fractional Fourier transform(Fr FT)is used to transform the variable speed signal into multiple order domains;Then,hierarchical decomposition is performed on the signals of each order domain,and finally the range entropy at each node is calculated.By applying the proposed method to the analysis of Gaussian white noise signals,1/f signals and experimental data sets,it is fully proved that the Fr HRE method has high stability and can extract signal features comprehensively.Aiming at the problem that random forest model cannot realize accurate recognition due to artificial parameter setting,this paper proposes to use HPO algorithm to find the optimal value of important parameters of random forest adaptively,and input the optimal parameters to obtain the optimized RF model.By comparing the identification results of different optimization models,it is verified that HPO-RF model has strong learning and identification capabilities.In addition,this article proposes a complete intelligent identification scheme for small faults in variable speed bearings based on HPO-VMDATVMF,Fr HRE,and HPO-RF,and further proves through experiments that HPO-VMDATVMF and Fr HRE have a positive impact on improving the recognition accuracy of the proposed model.