电光与控制, 2016, 23 (9): 90, 网络出版: 2021-01-26
融合寿命数据与退化数据的剩余寿命估计方法
Remaining Useful Lifetime Estimation by Combining Lifetime Data with Degradation Data
寿命数据 退化数据 扩散过程 极大似然估计 Bayesian方法 lifetime data degradation data diffusion process Maximum Likelihood Estimation(MLE) Bayesian method
摘要
剩余寿命(RUL)估计是设备健康管理的重要环节。基于扩散过程提出了一种融合寿命数据与退化数据的剩余寿命估计方法,利用首达时间的概念得到了剩余寿命的解析概率分布,并且给出了一种离线优化、在线更新的退化模型参数更新方法。首先将基于寿命数据的极大似然参数估计值作为Bayesian更新的初始值,然后通过Bayesian方法融合设备自身的退化数据更新退化模型的参数,最终实现剩余寿命的实时估计。实验结果验证了本文方法的有效性和优越性。
Abstract
Remaining Useful Lifetime (RUL) estimation is a significant part of health management. A RUL estimation method is proposed by combining the lifetime data with the degradation data based on diffusion process, and the RUL distribution is derived with the concept of the first passage time. In addition, the parameters are optimized offline and updated online. Firstly, maximum likelihood parameter estimation value based on the lifetime data is considered as the initial value of Bayesian updating. Secondly, the lifetime data and the degradation data are combined by Bayesian method to update parameters of the degradation model. Finally, the real time estimation of RUL is realized. The experimental results verify the effectiveness and superiority of the proposed method.
裴洪, 胡昌华, 司小胜, 张正新, 周绍华. 融合寿命数据与退化数据的剩余寿命估计方法[J]. 电光与控制, 2016, 23(9): 90. PEI Hong, HU Chang-hua, SI Xiao-sheng, ZHANG Zheng-xin, ZHOU Shao-hua. Remaining Useful Lifetime Estimation by Combining Lifetime Data with Degradation Data[J]. Electronics Optics & Control, 2016, 23(9): 90.