| Rolling bearing is an important basic part of the rotating machinery equipment in the industry production. The running status of rolling bearing affects the performance and safety of machinery equipment directly. Rolling bearing will experience a series of performance degradation states from the beginning to complete failure during the whole life cycle. The rolling bearing performance degradation assessment is aimed to describe the degradation degree in the whole life cycle.Therefore, it is very important to assess the performance degradation state of rolling bearing, in order to ensure the equipment working efficiency and prevent the happening of sudden accident.The whole life cycle vibration signal contains the useful information of rolling bearing performance degradation degree. In order to extract more comprehensive feature information, several time domain statistical indexes and frequency domain statistical indexes are used to extract time domain and frequency domain features of rolling bearing; the wavelet packet decomposition method is used effectively to extract time-frequency features of rolling bearing. And the rolling bearing fault feature set is established.As for the recognition of rolling bearing performance degradation states, the classification rules improved hypersphere support vector machine theory has been studied intensively in this thesis. Aiming at the uneven distribution characteristics of rolling bearing life cycle vibration signal, the hypersphere support vector machine is optimized by multi-kernel method, which multi-kernel hypersphere support vector machine model is constructed. The effectiveness of the optimized method is verified via the standard UCI data sets.In order to eliminate the blindness of artificial selection for support vector machine parameters, some intelligent optimization algorithms are studied in this thesis, such as genetic algorithms, particle swarm optimization, fruit fly optimization algorithm and chaos fruit fly optimization algorithm. The performance of each algorithm is analyzed in detail through basic function optimization experiment.Finally, the chaos fruit fly optimization algorithm is applied to the parameters selection process of multi-kernel hypersphere support vector machine, the CFOA-MKHSVM model is constructed. And an assessment index is proposed to quantitative evaluate the performance degradation degree for the rolling bearing based on normalized difference coefficient. |
