Abstract: Periodic impulses in vibration signals are useful to the detection of faults in axial piston pumps. However, in the working condition, the vibration signals of axial piston pump are often contaminated by heavy background noises and natural periodic impulses caused by the reciprocating movement of pistons. Therefore, extracting fault features is one of the most difficult tasks to identify faults in axial piston pumps. To solve this problem, the Teager energy operator(TEO) demodulation using improved clustering-based segmentation and L-Kurtosis method is proposed. Unlike the traditional clustering-based segmentation method, the improved version is a two-cycle one,it can extract the fault features out of the background noise and nature periodic impulse efficiently. L-Kurtosis is similar to kurtosis and easy to recognize impulses but is not like kurtosis to be sensitive to the outliers. The TEO demodulation is more suitable to extract faults than the traditional Hilbert demodulation, because the calculation of TEO is very simple. To illustrate the feasibility and performance of the present method, simulations and experimental data investigations are performed and the results are compared with the traditional clustering-based segmentation method. The results show that the proposed method enables the efficient detect cylinder fault and bearing fault in axial piston pumps.

Key words: axial piston pump, fault detection, improved clustering-based segmentation, L-kurtosis, Teager energy operator demodulation