Stochastic modeling and real-time prognostics for multi-component systems with degradation rate interactions

Many conventional models that characterize the reliability of multi-component systems are developed on the premise that component failures in a system are independent. By contrast, this article offers a unique perspective on modeling component interdependencies and predicting their residual lifetimes. Specifically, the article provides a stochastic modeling framework for characterizing interactions among the degradation processes of interdependent components of a given system. This is achieved by modeling the behaviors of condition-/degradation-based sensor signals that are associated with each component. The proposed model is also used to estimate the residual lifetime distributions of each component. In addition, a Bayesian framework is used to update the predicted residual lifetime distributions using sensor signals that are correlated with the real-time dynamics associated with the interactions. The robustness and prediction accuracy of the methodology are investigated through a comprehensive simulation study that compares the performance of the proposed model to a counterpart benchmark that does not account for degradation interactions.