一种基于轴承剩余寿命预测的状态维修优化决策方法

Method of Condition Based Maintenance Optimizing Decision Based on Residual Life Prediction

针对以往研究中状态维修的关键环节,剩余寿命预测不能更新的问题,提出一种融合贝叶斯方法的神经网络退化预测模型,实现利用实时传感信号动态预测轴承的剩余寿命分布。检验结果表明,该模型对轴承的剩余寿命预测比较精确。基于更新的剩余寿命分布,建立了以费用率最小为目标的轴承状态维修优化决策模型,求解得到最优的轴承预防性更换时间。

Aiming at the problem of previous studies that the key link of condition based maintenanceresidual life prediction always keeps changeless,put forward is a neural net work degradation prediction model which combined Bayes method to predict the residual life distribution of bearing dynamically. The verification results indicate the veracity of the model. Based on the updated posterior residual life distribution,established is the model of minimum expense ratio to maintain and optimize a bearing with the solution to the optimal replacement time of the bearing.