新研发光电产品的剩余寿命自适应预测方法

Adaptive Remaining Useful Life Prediction Method for Newly Developed Photoelectric Products

针对新研发光电产品存在先验信息不足、缺乏历史数据等问题,提出一种基于期望最大化(EM)算法的剩余寿命(RUL)自适应预测方法。然而,现有大多数RUL预测方法中普遍存在两个方面问题:在退化建模中存在一个潜在的假设,即当前时刻估计的随机参数与上一时刻随机参数的后验估计完全相等;在参数估计中假定存在多组同类型光电产品的历史退化数据,用于离线确定模型初始参数,致使RUL预测的精度受限于数据量。鉴于此,在状态空间模型的框架下构造一个新的退化模型,进一步推导出RUL分布的解析解;提出一种基于EM算法的自适应参数估计方法,以克服先验信息不足、缺乏历史数据等问题;通过GaAs激光器和光纤陀螺的实际退化数据进行实验研究。结果表明本文方法不仅可以提高RUL预测的精度,而且可以有效地应用于新研发的光电产品。

In this study, an adaptive remaining useful life(RUL) prediction method is developed based on the expectation maximization(EM) algorithm to solve the problems of insufficient prior information and lack of historical data with respect to the newly developed photoelectric products. Currently, two common problems are associated with majority of the existing RUL prediction methods. First, there is an underlying assumption in degradation modeling that a random parameter estimated at the current time is exactly equal to the posterior estimation of the random parameter at the previous time. Second, the historical degradation data are assumed to be available for parameter estimation based on which the initial model parameters can be determined for multiple photoelectric products of the same type. The RUL prediction accuracy is limited by data availability. Herein, we construct a novel degradation model under the state space model framework and derive the analytical form of the RUL distribution. Subsequently, we propose an adaptive parameter prediction method based on the EM algorithm to overcome the problems of insufficient prior information and lack of historical data. Finally, we conduct an experimental study with respect to the actual degradation data of a GaAs laser and fiber-optic gyroscope to denote that the proposed method improves the RUL prediction accuracy and can be effectively applied to the newly developed photoelectric products.