An adaptive decision method using structure feature analysis on dynamic fault propagation model

Effective online maintenance decisions for the troubleshooting of complex systems can avoid fault progressions and reduce potential losses. Inspired by the inhomogeneous topological nature of complex networks, in this study we intend to explore the pivotal node with high failure pervasion ability, so as to formulate an adaptive decision-making method. Dynamic Uncertain Causality Graph is introduced as the fault propagation model for evolving causalities. In order to evaluate the inherent topological structure feature of failure mode, the between centrality and non-symmetrical entropy are incorporated in the fault spreading risk measurement of nodes. Benefiting from solutions of time-varying structure decomposition and causality reduction on fault propagation model, the decision-making algorithm based on local causality structure achieves globally high efficiency and scalability. Verification experiments using generator faults of a nuclear power plant indicate the feasibility of this method in large-scale industrial applications.