Abstract: The safety analysis of complex industrial system is studied from the complex network viewpoint. Considering the complex electromechanical system as an integrated unit, the effect of the system’s topological characteristics on the prevalence of failure is analyzed by using the small world net theory, and the rules, which are capable of describing the safe sensitivity of the network nodes, are proposed. Subsequently, the failure grade diffusion model based on the small world clustering is established and the failure propagation intensity between the nodes is determined. The failure propagation paths with high risk and corresponding crisp nodes are found with the ant colony algorithm. Finally a durene production system is provided to illustrate the feasibility of the proposed method. The result shows that the proposed method can find the system crisp nodes efficiently, and can be used as a support next in risk assessment and maintenance decision.

Key words: Ant colony algorithm, Failure diffusion, Propagation intensity, Safety analysis, Small world net