不确定退化测量数据下的剩余寿命估计

剩余寿命估计是工程系统预测与健康管理的关键.目前,基于观测的系统退化数据进行剩余寿命估计得到了很大的关注.由于系统随机退化过程和测量误差的影响,测量数据中不可避免包含退化随机性和测量不确定性.然而,现有基于观测数据的剩余寿命估计研究中,没有将退化随机性和测量不确定性对估计的剩余寿命分布的影响同时考虑.鉴于此,提出了一种基于Wiener过程且同时考虑随机退化和不确定测量的退化建模方法,利用Kalman滤波技术,实现了潜在退化状态的实时估计.在退化状态估计的基础上,得到了同时考虑退化状态不确定性和测量不确定性的解析剩余寿命分布.此外,提出了一种基于极大似然方法的退化模型参数估计方法.最后,通过陀螺仪的退化测量数据验证了本文提出的方法优于不考虑测量不确定性的方法,可以提高剩余寿命估计的准确性.

Estimating Remaining Useful Life Under Uncertain Degradation Measurements

Remaining useful lifetime (RUL) estimation is a key issue in prognosis and health management for industrial systems.Currently,the use of the observed degradation data of a system holds promise to estimate its RUL.Due to the effect of system's stochastic deterioration and uncertain measurements,the measured data are inevitably contaminated by the stochasticity of the degradation and measurement uncertainty.However,in current studies of the RUL estimation based on the measured data,there is no report considering the effect of the degradation stochasticity and measurement uncertainty on the estimated RUL distribution.In this paper,a new degradation modeling approach is proposed based on Wiener process,which considers system's stochastic deterioration and uncertain measurements simultaneously,and the Kalman filtering technique is utilized to estimate the underlying degradation state.On the basis of the estimated degradation state,the analytical RUL distribution is derived which accounts for the uncertainties in the estimated degradation state and measurements.Additionally,a parameter estimation method for the developed model is presented based on the maximum likelihood method.Finally,a case study for gyros verifies the proposed method and the results indicate that the proposed method is superior to the method without considering uncertain measurements and can improve the accuracy of the estimated RUL.