Nonparametric nonlinear profile monitoring method for machine condition monitoring

Machine condition monitoring (MCM) aims to evaluate machine health conditions by statistically analysing data. Health indicators (HIs) such as Gini index, negative entropy, spectral kurtosis et al. have been commonly applied in MCM. But these HIs lack a statistical threshold and monotonic degradation tendency. To monitor the of rotating machines, nonparametric nonlinear profile method for MCM is proposed this paper. And Hotelling T-square indicator (HTHI) with constructed B-spline modeling statistic. Experimental validation on public bearing dataset shows that HTHI can effectively detect early faults monotonically trends, which enriches methods domain.