Application of pattern spectrum to the trend analysis of fault in generator sets

Fault features based on pattern spectrum were presented for analyzing tiny changes of energy and waveform of generator sets vibration and reflecting the trend of early faults. After extracting the pattern spectrum of the vibration signal with the operation of multiscale mathematical morphology, the pattern spectrum clearance and skewness be obtained by the statistics of the pattern spectrum. Its result can well reflect the tiny changes of vibration signal. The pattern spectrum clearance and skewness can reflect the fault trend of the generator sets better than the traditional vibration amplitude and intensity.