Mission reliability prediction methods for board-level electronic equipment based on physics of failure and Bayesian networks

Compared with the traditional handbook-based reliability prediction method, the PoF-based method can provide more accurate prediction for a component. But for the boardlevel electronic product with redundant features, it is very difficult to predict its mission reliability. The challenge lies in the consideration of the interrelationship between those components´ PoF models in board. This paper presents a method to model the board as a system, RBD model is introduced to model the system reliability, and then Bayesian Networks (BNs) is proposed to solve the RBD model. As for the components, Monte Carlo simulation is intended to fit their life distribution curves based on the PoF Models of components. And the reliability of the board can be obtained with these curves used as the inputs. Meantime, BNs model provides the basis for diagnosing the fault during the mission, which aims to discover the weaknesses of the system and to guide the improvement of the system-design.