Development of crack induced impulse-based condition indicators for early tooth crack severity assessment

Abstract

Diagnosis of gearbox tooth cracks at an early stage is important to prevent catastrophic failures. The time synchronous average (TSA) of vibration signals of gearboxes with a tooth crack mainly consists of the gear meshing frequency (GMF) and its harmonics, the crack-related amplitude modulation and frequency modulation (AM-FM) and the crack induced impulse (CII). To conduct tooth crack diagnosis, many diagnostic techniques have been developed based on the TSA signal. However, most of them achieved poor performance in terms of assessing tooth crack severity in the early crack stage. The main problem is that they are not developed using the signal components that can well indicate early tooth crack severity progression. To overcome this problem, this paper proposes a method to conduct a thorough analysis on the CII since it provides more valuable information on tooth cracks than other components. In the proposed method, the GMF and its harmonics and the crack-related AM-FM are first removed from the TSA signal using comb notch filters. The notch filtered signal mainly contains components related to the CII. A modal model is adopted to fit the dominant resonances in the notch filtered signal and modal parameters are obtained using the matrix pencil method. With the modal parameter estimates, the dominant resonances, which are single degree-of-freedom impulse responses (SDOF IRs), are identified and used to reconstruct the CII. Besides, the SDOF IRs are grouped according to their carrier frequencies. The energy of the reconstructed CII and the sum of the energy of the SDOF IRs with carrier frequencies in a specific frequency band are proposed as two new condition indicators (CIs) for tooth crack diagnosis. The effectiveness of the proposed method and the new CIs for early tooth crack severity assessment are demonstrated using simulated gearbox vibration signals and experimental gearbox vibration datasets.