A Wiener-process-inspired semi-stochastic filtering approach for prognostics

Semi-stochastic filtering method has been termed as one of the typical methods for predicting remaining useful life (RUL). Despite of its simple and direct framework, selecting or determining the conditional distribution of the condition monitoring (CM) measurements given the RUL is a challenging task remaining to be solved. To address this issue, this paper presents a novel Wiener-process-inspired semi-stochastic filtering approach for prognostics. First, the relationship between the degradation monitoring data modeled by Wiener process and the RUL of the system is explored. Inspired this relation, the conditional distribution of the CM measurements give the RUL is directly established and used as the observation equation in semi-stochastic filtering based prognosis method, and the parameters in Wiener process and semi-stochastic filtering model are estimated by the maximum likelihood estimation method. Then, the RUL of the concerned in-service system can be predicted with a probabilistic distribution based on the constructed state equation and the estimated model parameters using the CM data to date. To do so, the proposed prognosis method has an ability to integrate the historical data and the real-time CM data of the in-service system. Finally, the developed method is validated by the wear data of milling cutters and Lithium-ion button cells.