Dynamic risk assessment and critical equipment identification for multi-state USV based on multi-agent model and importance measure

Dynamic risk assessment and critical equipment identification are essential for ensuring safe operation and maintenance of multi-state unmanned surface vehicle (USV) systems. However, existing methods exhibit significant limitations in addressing functional coupling effects, including difficulties in quantifying inter-functional coupling strength and overlooking interdependent relationships among equipment components. This paper proposes a dynamic risk assessment and critical equipment identification method for multi-state USV based on coupling risk importance measure (CRIM) and multi-agent modeling. The method establishes a three-layer distributed modeling architecture integrating Functional Resonance Analysis Method (FRAM) and multi-agent theory. Dempster-Shafer (D-S) evidence theory is employed to quantify functional coupling coefficients by treating the six FRAM relationship matrices as independent evidence sources. A comprehensive analysis framework integrates equipment state evolution, functional variation propagation, and system risk assessment into a unified dynamic risk assessment model. A critical equipment identification algorithm based on CRIM is subsequently developed. Case studies using USV under four mission scenarios demonstrate significant advantages of the proposed method.