基于迭代融合优化算法和WPHM的牵引电机轴承可靠性评估

Reliability Evaluation of Traction Motor Bearing Based onIterative Fusion Optimization Algorithm and WPHM

为了评估列车牵引电机轴承的运行可靠性,提出一种基于反向指数鲸鱼群(OEWOA)与粒子群(PSO)迭代融合的优化算法和威布尔比例故障率模型(WPHM)的评估方法。提取振动信号的时域、频域特征指标,利用主成分分析法(PCA)进行特征信息的融合,将融合后的特征指标作为WPHM的响应协变量进行可靠性评估。针对模型中多个参数直接求解困难的问题,利用反向学习策略(OBL)和指数收敛因子提高鲸鱼算法(WOA)的搜索能力,迭代融合PSO算法良好的收敛能力;通过OEWOA的探索指导PSO的迭代,避免PSO陷入局部最优,从而彻底探索搜索空间,同时使用PSO限制OEWOA的搜索,以便更快地将解收敛到全局最优值。应用该优化算法对WPHM中多个参数进行求解,实现牵引电机轴承运行可靠性的评估。结果表明,该方法与PSO、WOA等优化算法比较,具有更快的收敛速度和寻优能力。

In order to evaluate the operation reliability of train traction motor bearings,this paper presented an evaluation method based on the iterative fusion optimization algorithm composed of Opposition and Exponential Whale Optimization Algorithm(OEWOA)with Particle Swarm Optimization(PSO)and Weibull Proportional Hazards Model(WPHM).Firstly,the time domain and frequency domain characteristic indexes of vibration signals were extracted,and the Principal Components Analysis(PCA)was used to fuse the characteristic information.The fused characteristic indexes were used as the response covariates of WPHM for reliability evaluation.To deal with the difficulty of directly solving multiple parameters in the model,the Opposition-Based Learning(OBL)and exponential convergence factor were used to improve the search ability of WOA within good convergence ability of the PSO algorithm to avoid PSO falling into local optimization.The OEWOA was introduced in exploration phase and it enabled the WOA to guide PSO for better local optima avoidance and PSO restricted the WOA search mechanism during exploitation phase,thereby converging the solution faster to a global optimum value.The optimized algorithm was used to solve multiple parameters in WPHM to realize the effective evaluation of the operational reliability of the traction motor bearing.The results of bearing reliability evaluation examples show that the method has faster convergence speed and optimization capabilities,compared with PSO and WOA and other optimization algorithms.