Time-Frequency Analysis Based Motor Fault Detection Using Deconvolutive STFT Spectrogram
Motor fault signals typically are non-stationary, for which, the conventional Fourier transform algorithm can not satisfy the demand of fault signals extraction. Time-frequency analysis based motor fault detection algorithm can effectively identify rotor faults, such as dynamic eccentricity and the unbalanced rotor fault, by detecting the time-frequency components of the stator current signal, which has been an important signal processing method in motor fault detection. This paper proposes a Deconvolutive Short-Time Fourier Transform spectrogram (DSTFT) based motor fault detection method. Compared with the commonly used Short-Time Fourier Transform (STFT) and Wigner-Ville Distribution (WVD) algorithm, DSTFT based method shows better time-frequency concentration and cross-term suppression performance, thereby improving the accuracy of motor fault detection.