Abstract: Against the backdrop of intensifying maritime strategic competition and the profound transformation of maritime security paradigms, to address the emerging unmanned and clustered challenges in maritime security, this paper designs a collaborative decision-making framework for unmanned surface vehicle (USV) clusters in maritime security scenarios. The framework systematically constructs a complete process from threat assessment to mission execution: first, a multi-feature fusion model is employed to quantify target threat levels, and interception points are pre-calculated using the parallel approach method; then, for resource-constrained situations, a K-means algorithm incorporating outlier influence is applied to cluster and consolidate interception points effectively; in path planning, an improved particle swarm optimization algorithm with inertia decay and bow-direction constraints is utilized to generate smooth trajectories; afterwards, a market-based mechanism is adopted for task allocation, achieving dynamic negotiation and resource coordination. Finally, simulation experiments under typical maritime security scenarios are conducted; the results show that effective interception can be accomplished in both resource-sufficient and resource-deficient settings.