Reliable fault diagnosis method using ensemble fuzzy ARTMAP based on improved Bayesian belief method

In this paper, a fuzzy ARTMAP (FAM) ensemble approach based on the improved Bayesian belief method is presented and applied to the fault diagnosis of rolling element bearings. First, by the statistical method, continuous Morlet wavelet analysis method and time series analysis method many features are extracted from the vibration signals to depict the information about the bearings. Second, with the modified distance discriminant technique some salient and sensitive features are selected. Finally, the optimal features are input into a committee of FAMs in different sequence, the output from these FAMs is combined and the combined decision is derived by the improved Bayesian belief method. The experiment results show that the proposed FAMs ensemble can reliably diagnose different fault conditions including different categories and severities, and has a better diagnosis performance compared with single FAM.