Adaptive redundant second generation wavelet design and gearbox fault feature extraction

To extract fault feature of gearbox signal corrupted by noise, a novel method to design adaptive redundant second generation wavelet (ARSGW) is developed, where the data-based optimization algorithm is adopted to design the initial prediction operator and update operator at each decomposition level, the initial prediction operator and update operator are interpolated with zero, and the redundant prediction operator and redundant update operator are gained. Splitting operation is unnecessary for ARSGW, and the approximation signal at each level is predicted and updated directly to retain the signal characteristics in time domain. The demodulation waveform and periodic impulses from the rub fault vibration signal of a large air compressor gearbox are desirably extracted with ARSGW, the approximation signal is demodulated, and the rotating frequency of high speed axis is regarded as the modulation frequency. Engineering results confirm that ARSGW outperforms the other wavelet.