A degradation-modeling based prognostic approach for systems with switching operating process

Abstract

Degradation-modeling based prognostic approach has been proved as an effective alternative to the conventional lifetime-data dependent residual life prediction method, and thus draw much attention of both scholars and engineers in the field of reliability. The degradation process of a system is the result of interaction between its inner states and working environments. To provide a reasonable reference for the sequential decision making based on prognostic result, the influence of operation process has to been incorporated into degradation modeling and prognosis. Therefore, this paper concerns the residual life prediction issue for system experiencing switching operation process whose influence on the system's performance degradation includes both deterioration and shocks. Besides the fact that the concerned system exhibits different deteriorating rates in each operation state, the change of operation states introduces external stresses and causes mutation in performance of the system. Therefore, the operation process depicted through a continuous time Markov chain (CTMC) is incorporated into the system's degradation modeling, based on which the system's residual life distribution is derived approximately yet explicitly after it is defined under the concept of first hitting time (FHT). Such a residual lifetime distribution is quite desired in prognostics and health management, especially for cases where online updating is required. The proposed approach is illustrated and validated by a numerical study.