Comparisons of Different Importance Measures with Hierarchical System Markov Bayesian Network Model

Importance measures are widely used in reliability analysis to identify weak components in the whole system and support maintenance decisions. In this paper, three different importance measures, namely Griffith importance measure (GIM), Wu importance measure (Wu IM) and integrated importance measure (IIM), are utilized to analyze weak components in hierarchical systems based on the hierarchical system Markov Bayesian network (HSMBN). First, HSMBN is conducted to describe the hierarchical system considering the dependent relationship among components and subsystems. The discrete-state Markov process model is then conducted to describe component or system degradation processes. A reliability information aggregation method is proposed based on the HSMBN under a Bayesian framework. The three importance measures mentioned above are then introduced and incorporated into the HSMBN. Finally, a case study on an aircraft electro-hydrostatic actuator (EHA) is used to verify the proposed model. The results show that HSMBN provides a comprehensive system model to analyze the dependent relationships among components and that different importance measures based on HSMBN are effective for identifying key states affecting the system performance.