Aim at the problem of fault diagnosis of roller bearings with rotating speed fluctuation

a method for the fault diagnosis of roller bearings was proposed by combining the chirplet path pursuit algorithm with the order cyclostationary demodulation method. In this proposed method

the chirplet path pursuit algorithm was used to obtain the bearing fault characteristic frequency. According to the bearing fault characteristic frequency

the envelope of the time domain vibration signal with rotating speed fluctuation was resampled at constant angle increments in angular domain

and the obtained angular domain steady signal was dealt by cyclostationary demodulation. The slice demodulation spectrum was then acquired and used to diagnose the fault of roller bearings. Simulation analysis and application examples show that the proposed method can exactly extract the outer and inner fault characteristics of roller bearings with rotating speed fluctuation. Compared with the envelope order spectrum method based on Wigner-Ville distribution

the proposed method can obtain more distinct fault characteristic of roller bearings.