Feature-based classifier ensembles for diagnosing multiple faults in rotating machinery
Applied Soft Computing
Fuzzy lattice classifier and its application to bearing fault diagnosis
Applied Soft Computing
| Hi-index | 0.00 |
Rolling element bearings are widely used to support rotating components of a machine. Due to close space locations of components in the machine, a vibration signal caused by bearing localized defects is easily overwhelmed by other strong vibration signals. Extracting the bearing fault signal from a multi-component signal mixture is thus significant to detect early bearing fault features and prevent machine breakdown. In this paper, a bearing fault diagnosis method, named cyclic spike detection method, is proposed to extract the weak bearing fault features from a multi-component signal mixture. Firstly, the optimal center frequency and bandwidth of a complex Morlet wavelet filter are determined by a simplex-simulated annealing algorithm along with a maximum sparsity objective function. The filtered signal is then obtained by applying the optimal wavelet filter to the multi-component signal mixture. After that, a new adaptive local maximum selection method is proposed to make the filtered signal succinct. Only a few spikes are retained to reveal potential cyclic intervals caused by bearing localized defects. Two multi-component signal mixtures, including a simulated signal and a real vibration signal collected from an industrial machine, are used to validate the effectiveness of the proposed cyclic spike detection method. The results demonstrate that the proposed method can extract the weak bearing fault features from other strong masking vibration signals and noise.