A Fault Identification Method Based on Multiresolution Generalized S-Transform

In order to detect fault impulse feature of rotating machinery in operation of non-uniform rotation, a fault identification method based on multiresolution generalized S-transform (MGST) is proposed in this paper. In this method, generalized S-transform (GST) with different scales is employed to generate time-frequency spectrums for the fault vibration signal. Then, normalize these spectrums and calculate the shift factors, phase factors, frequency factors and scale factors of the maximum energy, thus, an impulse component with the maximum energy could be extracted. Repeat this iteration for several times, all the impulse components could be extracted, combining the rotating speed information and structural parameters, the fault could be identified. The simulation proves that the proposed method expresses fast decomposition speed and high precision.