Automatic extraction of fault features for diesel engine based on bispectrum approach and variable precision rough set theory
Huimin Zhao, Jianmin Mei, Xiao Yunkui, Long Qiao, Xianglong Chen · Zhendong yu chongji · 2011
The unstable vibration signal of diesel engine crankshaft bearing was analyzed by using the method of bispectrum.According to the symmetrical property of bispectrum,the area below the diagonal in the first quadrant was divided equally into some identical regions in which the average magnitude of bispectrum components was calculated and taken as the feature of the analysed object.The variable precision rough set theory(VPRST) was then used to extract the feature parameters that correlate with the fault parts tightly.The analysis results show that the noise buried in the unstable vibration signal of crankshaft bearing can be eliminated by the bispectrum approach,the key factors relevant to the analysed object can be mined in accordance with VPRST,and the fault feature can be extracted automatically by bispectrum analysis combined with VPRST.