A New Decision-Diagram-Based Method for Efficient Analysis on Multistate Systems

Abstract

Multistate systems can model many practical systems in a wide range of real applications. A distinct characteristic of these systems is that the systems and their components may assume more than two levels of performance (or states), varying from perfect operation to complete failure. The nonbinary property of multistate systems and their components makes the analysis of multistate systems difficult. This paper proposes a new decision-diagram-based method, called multistate multivalued decision diagrams (MMDD), for the analysis of multistate systems with multistate components. Examples show how the MMDD models are generated and evaluated to obtain the system-state probabilities. The MMDD method is compared with the existing binary decision diagram (BDD)-based method. Empirical results show that the MMDD method can offer less computational complexity and simpler model evaluation algorithm than the BDD-based method.