Abstract: In order to extract the fault characteristic from vibration signal of a gearbox with variable rotate speed, the order cyclostationary demodulating approach based on chirplet path pursuit is proposed. In this proposed approach, the rotate speed signal is estimated by the chirplet path pursuit. According to the rotate speed signal, the angular domain cyclostationary signal is obtained by even angle sampling to the original signal. The autocorrelation function of angular domain signal is calculated and the slice of autocorrelation function on the characteristic cycle order is computed. Do the demodulating on the slice to acquire the slice demodulation spectrum which is used to diagnosis the gear fault. Since the chirplet path pursuit has the advantage of high precision and good anti-noise ability, moreover, the cyclostationary demodulating approach can extract the cycle fault feature from noise signal. So, the proposed approach inherit the merit of both, and is suitable for extracting fault feature from gear signal with variable rotate speed. Simulation and application examples show that the proposed approach is an effective way to extract the gear fault characteristic from vibration signal of gearbox with variable rotate speed.

Key words: Chirplet, Cyclostationary demodulating, Gear tooth fault diagnosis, Order tracking, aerodynamic noise, large eddy simulation, pantograph, pantograph fairing, reduction noise design, high-speed train