Research on Optimization Method for Intelligent C2 Systems Based on PREA-OODA Loops

To address the over-reliance on empirical decision-making in operational force grouping and the insufficient dynamic adaptability of command topology in intelligent Command and Control (C2) systems, this paper proposes an intelligent C2 system optimization method that integrates PREA-OODA dual-loop driving, constrained multi-objective optimization, and graph reinforcement learning. The method first establishes normalized models for collaborative planning complexity and command-and-control effectiveness, and formulates a multi-objective optimization problem by introducing hard constraints such as resource capacity and functional matching. A constraint-enhanced NSGA-Ⅲ grouping generation algorithm is designed, which guarantees the feasibility of solutions throughout the evolutionary process through feasible-solution seeding, deterministic repair operators, and a constraint dominance mechanism. An edge-level GNN-PPO graph reinforcement learning framework is constructed to achieve dynamic optimization of command topology, leveraging graph-based topology representation, an edge-level refined action space, and a complexity penalty mechanism. A PREA-OODA event-driven closed-loop mechanism is established to realize dual-layer linkage between global planning and local agile adjustment. Validation on a joint amphibious landing campaign demonstrates that the constraint-enhanced grouping algorithm can generate feasible solutions with zero constraint violation. After GNN-PPO optimization, the network response time is reduced by 12.0%, the survivability is improved by 29.7%, and the comprehensive effectiveness is increased by 11.1%, outperforming baseline methods. This research provides a referable methodological paradigm and technical path for the optimal design and effectiveness verification of intelligent command and control systems in joint operations.