Demodulation for hydraulic pump fault signals based on local mean decomposition and improved adaptive multiscale morphology analysis

Scales of IAMMA are adaptively determined by morphological features of signal, thus fault features of a hydraulic pump fault signal presented in multi-scales can be adaptively demodulated. In some coefficient range, IAMMA outperforms AMMA in demodulation ability based on the same SE, and it is less susceptible to noises than AMMA. The best performance of IAMMA with triangle SE is stronger than that of IAMMA with plat and semi-circle SE when they demodulate the same fault signal of hydraulic pump. Compared with traditional demodulation methods of HT and TKEO, IAMMA is adaptive and has stronger demodulation ability. An evaluation method based on kurtosis, power and standard deviation is proposed, by which some PFs which are rich in fault features can be selected as data source.