融合随机退化过程与失效率建模的设备剩余寿命预测方法

Remaining Lifetime Prediction by Integrating Stochastic Degradation Process with Hazard Rate

剩余寿命预测对于设备的维修与保养具有十分重要的意义。现有的剩余寿命预测方法大多只利用了设备的当前退化信息,对设备的历史寿命信息没有充分利用,而这些信息往往包含着设备寿命的演化信息,对于准确预测设备的剩余寿命具有重要意义。针对这个问题,提出了一种融合随机退化过程与失效率建模的设备剩余寿命预测方法。该方法首先将设备的退化过程建模为Wiener过程,然后利用Cox比例失效模型建模的方法融合设备退化过程对设备失效率的影响,由此达到利用设备历史监测信息的目的。进一步通过Bayes方法,利用当前退化监测信息对退化过程模型的参数进行更新,基于此进行剩余寿命预测,从而实现设备历史数据与当前数据的有效融合。最后,通过激光发生器的退化测量数据验证了提出的方法,说明该方法是有效的,具有一定的应用价值。

Remaining lifetime prediction is of vital importance in equipment maintenance and repairment.Most of the remaining lifetime prediction methods now available only use the current degradation information of equipment, but does not take full advantage of the historical lifetime information, while this information always contains the evolution information of equipment lifetime, which is of great significance for accurate predicting of the remaining lifetime. To solve this problem, a remaining lifetime prediction method is proposed by integrating the stochastic degradation process with hazard rate. Firstly, the equipment degradation process is modeled as a Wiener process, and then Cox proportional hazard model is used for modeling, which integrates the influence of equipment degradation process on the hazard rate, thus attaining the purpose for using the historical monitoring information. Furthermore, parameters of the degradation model are updated with the current degradation monitoring information by Bayesian method, based on which, the remaining lifetime can be predicted, hence realizing effective fusion of equipment historical data and current data. Finally, the proposed method is verified by the degradation measurement data of laser generator. It is demonstrated that the proposed method is valid, with certain application value.