Classification of time-frequency representations using improved morphological pattern spectrum for engine fault diagnosis

Time-frequency representations (TFR) have been intensively employed for analyzing vibration signals in the field of mechanical faults diagnosis. However, in many applications, TFR are simply utilized as a visual aid. It is very attractive to investigate the utility of TFR for automatic classification of vibration signals. A key step for this work is to extract discriminative parameters from TFR as input feature vector for classifiers. This paper contributes to this ongoing investigation by developing an improved morphological pattern spectrum (IMPS) for feature extraction from TFR. The S transform, which combines the separate strengths of the short time Fourier transform and wavelet transforms, is chosen to perform the time-frequency analysis of vibration signals. Then, we present an improved morphological pattern spectrum (IMPS) scheme, which utilizes the first moment replace of the volume measure used in traditional morphological pattern spectrum (MPS) method. The promise of IMPS is illustrated by performing our procedure on vibration signals measured from an engine with five operating states. Experimental results have demonstrated the presented IMPS to be an effective approach for characterizing the TFR of vibration signals in engine fault diagnosis.