| As one of the main components of rotating machinery,the state of rolling bearing has a direct effect on the performance of the whole machine.In actual engineering,the inner ring,outer ring and rolling body of the bearing and other parts will gradually appear failure,rapid and accurate identification of the cause of bearing failure,and take the corresponding maintenance methods,to maintain the smooth operation of the machine and avoid causing major accidents have key practical significance.This paper takes rolling bearings as the research object,and studies the extraction of composite fault characteristics of rolling bearings as well as intelligent fault diagnosis methods,the main work content is as follows:(1)A method for extracting fault features of rolling bearing based on ICEEMDAN is proposed.By introducing the cubic trigonometric cardinal spline interpolation function instead of the cubic spline interpolation function in the original algorithm,the problem of over-envelope and under-envelope in the CEEMDAN algorithm is solved,and the error generated by the decomposition is reduced.The experimental results show that ICEEMDAN can extract the fault features better than CEEMDAN in the single fault of the bearing.(2)A method for extracting composite fault features of rolling bearings with improved FastICA’s single-channel blind source separation is proposed.The ICEEMDAN method is decomposing signal to construct a multi-channel signal,estimate the number of source signals based on the SVD theory,and construct the CSI integrated signal index based on the kurtosis,correlation coefficient and root mean square to select the optimal observed signal as the input for blind source separation.In the process of FastICA improvement,the convergence stability is enhanced by the most rapid descent method in order to overcome the drawback that the original algorithm is vulnerable to the initial value.The results show that the proposed method can clearly extract the 4-times frequencies of the characteristic frequencies of the outer ring and rolling body of the bearing,and realize the accurate diagnosis of the composite bearing fault(3)An intelligent fault diagnosis method of improved grey wolf algorithm(IGWO)optimized support vector machine(SVM)is proposed.Based on the processing of bearing vibration signals based on ICEEMDAN,the multiscale fuzzy entropy of each eigenmodal component is extracted as the feature vector input to the SVM model.An exponential function is introduced to optimize the convergence factor of the traditional gray wolf algorithm and combined with the dynamic mean position update strategy to propose the IGWO algorithm.The two important parameters c and g of the SVM are adopted to further improve the recognition accuracy and efficiency of the SVM.Experimental verification of different types of composite crack faults in rolling bearings and comparison of the recognition effect of other different classifiers,the results show that the method in this paper can well determine the different types of composite faults in bearings,the fault recognition accuracy reaches 99.38%,the recognition effect is better than other diagnostic methods. |
