UUV Swarm Threat Assessment via DBN Tracking, Vieta Ranking, and Distance Fusion

Unmanned Underwater Vehicle (UUV) swarms operating in complex marine environments must accurately assess threats from surrounding targets to ensure mission success and navigational safety. However, existing threat assessment methods face three fundamental bottlenecks when applied to underwater swarms: the inability to track temporally evolving target intentions, reliance on subjective indicator weighting for multi-target ranking, and vulnerability to spatially heterogeneous sonar noise. This paper proposes a hierarchical threat assessment framework that addresses these bottlenecks through three integrated modules. First, a Dynamic Bayesian Network with a specially designed heading factor tracks target intention over time, propagating threat probabilities across sequential observations and enabling early warning before the closest point of approach. Second, a Vieta’s theorem-based algebraic ranking algorithm constructs comprehensive threat vectors via elementary symmetric polynomials of six indicator utilities, avoiding explicit expert-defined weighting coefficients in the multi-attribute ranking stage while capturing both independent and synergistic indicator interactions. Third, a distance-weighted swarm aggregation strategy suppresses individual sonar noise by assigning higher fusion weights to geographically closer nodes, exploiting the spatial diversity inherent in swarm configurations. Simulation experiments under representative target-motion scenarios validate the framework across four complementary experimental studies. Results demonstrate that the DBN reduces output variance by over 56% compared to static Bayesian networks and responds to abrupt intention changes within 15 s. The algebraic ranking algorithm achieves identical prioritization to TOPSIS without requiring any manual or data-dependent weights. The distance-weighted aggregation reduces root mean square error by 63.2% and improves signal-to-noise ratio by 8.7 dB over equal-weight averaging. The proposed framework provides a principled and interpretable solution for simulation-based autonomous threat perception in representative underwater swarm scenarios.