Filter characteristics of variational mode decomposition method and its application in gearbox fault detection
                              DING Wenjie1, ZHAO Wuyun1, ZHANG Zheng2, WANG Changming2, DING Shuyong3

（1.College of Mechanical and Electrical Engineering，Gansu Agricultural University，Lanzhou 730070, China;2.College of Mechanical Engineering, Zhejiang University of Technology, Hangzhou 310014, China;3.College of Zhijiang, Zhejiang University of Technology, Shaoxing 312030, China）

Abstract: Aiming at the problem of better application of a new timefrequency analysis tool variational mode decomposition in fault diagnosis, the equivalent filtering characteristics of variational mode decomposition and the influence of different parameters on them were studied by generating different fractional Gauss noise, and the method of K value identification based on spectral envelope and central frequency initialization method were proposed. Firstly, the maximum value of the spectrum was extracted, and the spectral envelope was constructed by interpolation. Then, the maximum value of the spectrum envelope was detected, and the number of extreme points was used as the K value, and the abscissa of the extreme point was normalized as the central frequency spectrum of the initialization. The vibration signals of horizontal driving machine of automobile seat were tested. The results indicate that the K value recognition method based on spectral envelope effectively simplifies the iterative process of VMD and improves the computational efficiency of VMD, the decomposition result is more reasonable.
Key words: variational mode decomposition（VMD）; fractional Gauss noise; equivalent filter characteristic; spectral envelope; gearbox fault detection