A Novel Roller Bearing Condition Monitoring Method Based on RHLCD and FVPMCD

Effective condition monitoring provides some benefits such as improving safety and reliability. Roller bearing is the key component of rotating machinery, and a novel roller bearing condition monitoring method based on rational Hermite interpolation-local characteristic-scale decomposition (RHLCD) and fusion variable predictive model-based class discriminate method (FVPMCD) is proposed in this paper. RHLCD can adaptively decompose any complex signal into a sum of rational intrinsic scale components (RISCs), whose instantaneous frequency has physical meaning. In addition, targeting the limitation of variable predictive model-based class discriminate method (VPMCD), FVPMCD is presented. First, four kinds of common models are used to recognize a sample. Then, the recognition results of each model are satisfied, and the recognition probability of each state is calculated. Finally, the largest recognition probability of the state is chosen to recognize categories. The analytical results of experimental signals indicate that the proposed condition monitoring approach can identify the states of roller bearing effectively.