Maximum Correlated Kurtosis Deconvolution and Overlapping Group Shrinkage Denoising for Incipient Fault Diagnosis of Helicopter Planetary Gear Train

Abstract

Planetary gear train fault diagnosis is critical to Prognostics and Health Management of the helicopter. Generally, the localized gear fault will cause periodic impulses in the vibration signal. However, the fault characteristic generated by incipient fault is often weak and submerged in strong background noise, resulting in difficulty in fault feature extraction. To address this issue, a method combining Maximum Correlated Kurtosis Deconvolution(MCKD) and Overlapping Group Shrinkage(OGS) algorithms is proposed. Firstly, Fruit fly Optimization Algorithm(FOA) was employed to search for the optimal influencing parameters of MCKD, and the impulses of the raw fault signal could be enhanced after processed by MCKD adaptively. Then, the deconvolution signal was further processed by OGS to eliminate the residual noise. Finally, the fault characteristic frequency components could be identified by analyzing the envelope spectrum of the filtered signal, and achieved the incipient fault diagnosis of helicopter planetary gear. The proposed method is validated using the vibration signal of the planetary gear train in a helicopter transmission test rig, and a slight cracked planetary gear fault is successfully identified.