A New Improved Synchrosqueezing Transform Based on Adaptive Short Time Fourier Transform

The vibration signals during the run-up and run-down periods of rotating machinery are nonstationary. The normal types of time frequency (TF) analysis algorithms have been widely used in the nonstationary condition. But those methods show imperfection for drastically changing vibration signal. Synchrosqueezing Transform (SST) is an adaptive and invertible transform. However, the original SST method is not suitable for vibration signals with abruptly instantanous frequency (IF). In this paper, SST based on adaptive short time Fourier transform(ASTFT) is proposed to improve the time resolution of drastically changing vibration signal. Compared with Short Time Fourier Transform (STFT), SST and Generalized Synchrosqueezing Transform (GST) based on STFT, SST based on ASTFT is of good performance with high computing speed. Experiments were conducted by using SQI motor simulator to validate the proposed method. The vibration signals of a motor with fault were collected during the run-down process and analysed by the proposed method. The results showed that the fault feature frequency of the motor can be accurately captured, which suggests that the proposed approach is effective for analyzing the actute variation signals.