Gear defect detection through model-based wideband demodulation of vibrations

Amplitude and phase modulations in gear vibrations have been considered as ''gear fault signatures'' since gear faults strongly modulate a gear's vibrations. Previously, the modulation signals must be estimated from a bandpass-filtered gear vibration due to the limitation of the demodulation schemes. For better gear defect detection, a model-based demodulation scheme is developed in this paper to exploit the information contained in wideband gear vibrations. A gear vibration model that can include a large number of tooth meshing harmonics as modulation carrier is described. Based on this model, a wideband demodulation algorithm is established to derive amplitude and phase modulation signals from a gear's vibration signal average. No bandpass filtering is required. The effectiveness of this scheme is confirmed and compared to the state of the art using a vibration average of a gear with two defects of different sizes. (C) 1996 Academic Press Limited