Directional Markov Chain Monte Carlo Algorithm for Fast Dynamic Reliability Assessment

Abstract

Failure problems of many electromechanical devices are caused by the interaction of discrete disturbance and continuous degradation. A challenge presented by such multiscale failure behavior is how to implement fast dynamic reliability assessment. The Directional Markov Chain Monte Carlo (DMCMC) algorithm was thus presented to resolve the problem with variable steps. Thanks to the definition of failure space and the directional sampling principle, the computational cost was thus reduced. Simulation of a servo valve case demonstrated its efficiency.