Compound faults detection of rotating machinery using improved adaptive redundant lifting multiwavelet

Due to the character of diversity and complexity, the compound faults detection of rotating machinery under non-stationary operation turns into a challenging task. Multiwavelet with two or more base functions and many excellent properties provides a possibility to detect and extract all the features of compound faults at one time. However, the fixed basis functions independent of the vibration signal may decrease the accuracy of fault detection. Moreover, the decomposition result of discrete multiwavelet transform does not possess time invariance, which is harmful to extract the feature of periodical impulses. To overcome these deficiencies, based on the Hermite splines interpolation, taking the minimum envelope spectrum entropy as the optimization objective, adaptive redundant lifting multiwavelet is developed. Additionally, in order to eliminate error propagation of decomposition results, adaptive redundant lifting multiwavelet is improved by adding the normalization factors. As an effective method, Hilbert transform demodulation analysis is used to extract the fault feature from the high frequency modulation signal. The proposed method incorporating improved adaptive redundant lifting multiwavelet (IARLM) with Hilbert transform demodulation analysis is applied to compound faults detection for the simulation experiment, rolling element bearing test bench and traveling unit of electric locomotive. Compared with some other fault detection methods, the results show the superior effectiveness and reliability on the compound faults detection.