Study On Signal Extraction And Fault Identification Of Rolling Bearing Of Vibration Motor Based On Wavelet-EEMD- BSS And Modified KNN Algorithm

With the rapid development of science and technology, the use of vibration machinery develop towards large-scale, intelligent. Therefore, the stability of vibration machinery also requires more and more high, more and more attention are paid to the condition monitoring and fault diagnosis.The vibration source is the power source of vibrating screen, the failure of vibrating source will result in a shaker to stop running, even cause an accident, cause the enormous economic losses. So, it is very important significance to improve the service life of the vibration sieve and the production efficiency of the source of vibration fault diagnosis and prevention, and it plays an important role in promoting the development of fault diagnosis technology of the vibrating screen. In this paper, on fault identification and signal extraction of rolling bearing of vibration motor based on EEMD- BSS and modified KNN algorithm.The difference of wavelet and Fourier transform are analyzed, the wavelet transform is analyzed, and the wavelet functions and the basic theory of wavelet packet are analyzed. The simulation of extracting weak signal is verified that wavelet is effective in weak signal extraction. Through comparing effects of a test signal denoising using wavelet and using Butterworth, it proves that wavelet has more advantages than the Butterworth.The basic concepts and the principle of EMD and EEMD algorithm are introduced. Through the comparative analysis of the analog signal and the comparative analysis of experimental data from actual conditions in the EEMD algorithm, it is proved the ability of the EEMD algorithm against aliasing effect is obviously better than the ability of the EMD algorithm. According to the advantages of EEMD algorithm aliasing effect against the blind source separation, combination of traditional BSS algorithm, we propose an improved EEMD-BSS algorithm. This algorithm can separate signals when the number of observed signals less than the nember of vibration source. The validity of the EEMD-BSS algorithm is proved through the simulation.After analyzing the mechanism of rolling bearing of vibration motor, calculating and analyzing the frequency of the signal of rolling bearing fault. Modeling and simulation the rolling bearing of the vibration motor by using Pro/E and Adams software. According to the EEMD—BSS algorithm is proposed in this paper, the simulating fault signals is decomposed and restructured by wavelet packet, then the restructured signal is decomposed by EEMD.The original signal and the 7~10th order components are restructured. The restructured signal is separated by BSS algorithm. The results of separation shows that the EEMD-BSS algorithm can extract effectively the fault feature frequency from the signal of the single channel. It is cope with the underdetermined problem into a well posed problem. So, it lays the foundation for the blind separation of fault signals.Aiming at the shortcomings of traditional KNN algorithm, this paper proposed a modified weighted KNN algorithm, the weighting coefficients are formed by feature attributes based on the accuracy of the classification. This weighted KNN algorithm is used in the classification of the rolling bearing samples of vibration motor. Comparison of classification results of the traditional KNN algorithm and the results of the modified weighted KNN algorithm, it is proved that the modified weighted KNN algorithms can effectively improve the classification accuracy. Aiming at the deficiency of traditional KNN algorithm, another modified KNN algorithm is proposed based on asymmetric proximity function. The rolling bearing samples of the vibration motor were classified by the modified KNN algorithm. The results of classification shows that the modified KNN algorithm based on asymmetric proximity function can classify the samples more accuracy than the modified weighted KNN algorithm.Studying on fault diagnosis technique of rolling bearing of vibration motor based on EEMD- BSS and modified KNN algorithm in this paper, which has reliable theoretical basis. It will provide a reference to the research of the fault diagnosis of vibration machinery.