Research And System Development Of Rolling Bearing Fault Diagnosis

With the development of the society and economy,the problem of the health management of equipment has penetrated into all walks of life,and the automatic and multi-functional modern large-scale mechanical equipment constructed from small mechanical parts can be seen everywhere,so some of the parts that play a key role in the normal operation of the whole equipment are particularly important.The rolling bearing is not only the key supporting part of the aero engine,but also widely used in the fields of machine tools,high-speed rail,automobile,and household electric appliances.The real-time online monitoring of these key parts and the timely diagnosis of their fault types can reduce some of the unnecessary losses,ensure the normal operation of the whole equipment and improve the utilization rate of the equipment.This paper aims at developing a reasonable and available fault diagnosis platform for rolling bearings.first of all,a new method of rolling bearing fault diagnosis based on CEEMDAN-MPE-SVM is proposed,then through the concrete test to prove that the proposed method also has good practicability in the aspect of fault prediction,finally,the fault diagnosis system of rolling bearing is developed based on the GUI of MATLAB,and verify the fault diagnosis system is reasonable and effective by practical data.The specific research contents include:(1)The fault diagnosis method of rolling bearing based on CEEMDAN-MPE-SVM is proposed.From the three aspects of signal preprocessing,feature extraction and fault identification,first,the rolling bearing fault data obtained from the rolling bearing fault simulation test rig of the Electrical Engineering Laboratory of Case Western Reserve University are used,de-noising using wavelet threshold for data,a total of nine denoising methods are listed,through the vibration signal of the bearing outer ring?the inner ring and the rolling body,the comparison and analysis of the nine denoising methods are made by using the energy ratio,similarity and signal-to-noise ratio of the three signal denoising effect evaluation indexes,the optimal denoising method is obtained finally;Then,the five modal decomposition methods of EMD? EEMD? CEEMD? MEEMD and CEEMDAN of the simulation signal are filtered out by calculating the time-consuming,orthogonality and false components,and a reasonable modal decomposition method is filtered out,and the signal is reconstructed by using the rule of interrelation number;the multi-scale permutation entropy of 0.5-1s,1.5-2s,2.5-3s and 3.5-4s of each working state are calculated,and which can be seen as the eigenvectors,the trend of multi-scale entropy value(MPE)of the four sets of signals is approximately the same under the same condition;Finally,a total of 52 groups of samples were collected,48 groups as training set,4 groups as test set,input SVM,GRNN and PNN to identify the accuracy rate respectively,and the final recognition rate of SVM is 100%.Therefore,the fault identification method of rolling bearing based on CEEMDAN-MPE-SVM is proposed.(2)The practicability of CEEMDAN-MPE method for bearing cage fault prediction is verified in the boring machine bearing of the connecting rod mechanism.One accelerometer are arranged on 4 spindles respectively,the signal is connected to the computer by the acquisition card to record the acceleration of 4 bearings in real time,after data processing,CEENDAN decomposition and MPE value calculation,the failure of the bearing cage is predicted in advance,finally,it is consistent with the actual results,in this way,the machining precision of the inner hole of the connecting rod can be guaranteed and the utilization rate of the equipment can be improved.(3)A fault diagnosis system for rolling bearings is developed based on MATLAB GUI.Based on the theoretical basis of CEEMDAN-MPE-SVM method and GUI for technical support,a set of fault diagnosis system which can operate in real time without backstage programming is developed.The system consists of three modules: the user landing module,the fault feature extraction module of the rolling bearing and the fault diagnosis module,the functions of each module are realized through the layout of the interface,the editing of the control properties and the debugging of the M file.The system can output the corresponding fault type after input the original signal and several simple mouse operations.Finally,the experimental data of the first two chapters are input into the system,and the fault type obtained is in agreement with the actual experimental results,which effectively verifies the rationality of the system.