Method of assessing the multi-state of a rolling bearing based on CFOA-HSVM two measures combination

To assess the multi-state of a rolling bearing more effectively and simultaneously, a unified assessment method is proposed based on chaos fruit fly optimization algorithm hyper-sphere support vector machine (CFOA-HSVM) two measures combination. Aiming to the blindness of parameters selection for HSVM, multiple parameters of HSVM can be searched the optimal values using chaos theory combined with fruit fly optimization algorithm (CFOA), and HSVM model based on CFOA can be constructed. On above basis, the minimum of difference coefficient, that is generalized minimum distance, can be extracted and regarded as geometric assessment distance measure that is sensitive to absolute value. Meanwhile, angle cosine distance is introduced to calculate angle cosine distance between classification state and normal state, and is regarded as assessment distance measure which is sensitive to the direction. Then the assessment measure of generalized minimum distance is compensated and the unified assessment index of multi-state of the rolling bearing is constructed based on two measures combination, further the assessment model can be constructed and the assessment curve can be obtained. Experiments show that the proposed method can assess the different fault location and different failure degree of the rolling bearing more effectively and simultaneously.