Early Detection of Gear Crack Based on Cyclic Time Series

On the basis of gear vibration signals analysis,one novel method of detecting faults based on linear periodical time-varying autoregressive (LPTV-AR) model according to its cyclostationarity was proposed.The algorithm of identifying model parameters using higher-order cyclic-cumulants was established,which possessed the ability to suppress additive stationary noise.The proposed method was used to detect and predict the incipient gear crack fault of a helicopter.The results show that the approach has good performance of prediction and can detect early gear crack fault using the kurtosis of residual signals.At the same time it provides an idea of detecting faults of rotating machinery using only normal samples.