Demarcation of Gear Defect from Roller Bearing Defect Based on Optimized Wavelet Decomposition Technique

This article is concerned with the vibration-based analyses to differentiate a combination of faulty gear signal with strong interfering bearing signals. Measured signals are collected from a vehicle gearbox of a midsize passenger car. The main objective aims to identify the faulty gear signal from the bearing signal. The proposed method utilizes an optimized wavelet decomposition technique to decompose the experimental vibration signals into monocomponents. Daubechies wavelet function (db02) is implemented in the decomposition process, and optimum/minimum entropy (E) is used to select the optimum level of the best wavelet decomposition tree. The decomposed and optimized signal is illustrated in the form of Power Spectral Density (PSD), which is a widely used method for the nonstationary signals to overcome the limitations of Fast Fourier Transforms (FFT) for nonlinear signals. An experimental study is conducted on the investigated gearbox with localized artificial defects: one in the pinion tooth on the secondary shaft and one in the roller element of the bearing on the input shaft. The results show that the localized defect in a pinion can be effectively distinct from the bearing element defect and diagnosed with the help of the proposed technique.