LAUSR

Planetary gearbox (PGB) usually work in harsh working conditions with low speed and heavy load, and they are prone to wear. Different from the local faults, the distributed faults such as tooth surface wear are often weak and difficult to detect in the early stage, and it is difficult to extract fault characteristic. This paper presents an early fault diagnosis method for the distributed tooth surface wear of PGB to solve this problem. The proposed multi-channel optimal maximum correlation kurtosis deconvolution (MCO_MCKD) algorithm is used to extract fault characteristic. In order to enhance the effect of fault characteristic extraction (FCE), the algorithm first uses the sliding window principle to segment the input signal to establishes multiple channels for maximum correlation kurtosis (max_CK) optimization based on all the short signals obtained. The finite impulse response (FIR) filter with the max_CK is selected to filter the input signal, in order to realize FCE. The influence of tooth wear is mainly reflected in the frequency-domain signal amplitude. In order to realize early fault diagnosis, the frequency-domain statistical indicator fault characteristic energy ratio (FCER) is proposed based on this characteristic. The health status of the equipment is monitored by calculating the FCER of the signal after FCE. Early fault diagnosis is realized based on the mutation of the FCER. The simulation results show that MCO_MCKD algorithm has strong robustness. The experimental results show this proposed method is effective and superior.