Adaptive MOMEDA based on improved advance-retreat algorithm for fault features extraction of axial piston pump

The fault information of axial piston pump bearings is inevitably submerged by violent natural periodic impulses. Therefore, an accurate extraction of fault impulses remains a challenging problem. A hybrid method of MOMEDA and TEO is proposed to extract periodic impulses in this study. Firstly, the deconvolution periods of multiple periodic components in the original vibration signals are analysed using Kurtosis. Then, an advance-retreat algorithm is used to optimize the filter length of MOMEDA. After multiple input parameters are determined adaptively, the MOMEDA is used to enhance the various periodic impulses respectively. Finally, TEO demodulation is employed to further obtain fault frequencies. Experimental vibration data is used to verify the advantages of this method for periodic impulses extraction. The results are then compared with traditional deconvolution and decomposition techniques to prove the superior performance of the proposed approach in terms of its better accuracy and reduced processing time.(c) 2021 ISA. Published by Elsevier Ltd. All rights reserved.

Automated summary

A hybrid method of MOMEDA and TEO is proposed to extract periodic impulses in axial piston pump bearings. The deconvolution periods and filter length are determined using Kurtosis and an advance-retreat algorithm. MOMEDA is used to enhance the periodic impulses, and TEO demodulation is employed to obtain fault frequencies.