An Integrated Dispatching and Centralized Control Coordinated Operation Method for Large-Scale Offshore Wind Power Clusters

Abstract

To achieve accurate dispatching and efficient coordination of large-scale offshore wind farms, this paper proposes a three-level integrated operation method integrating dispatching, centralized control, and wind turbines. Aiming at the low utilization of regulation capability, slow response, and weak fault tolerance of traditional dispatching modes, an information-physical fusion based three-level architecture is constructed. The dispatching layer adopts LSTM prediction and multi-objective particle swarm optimization for macroscopic task allocation. The farm-level control layer performs internal optimization using real-time constraint satisfaction and Kalman filtering. The turbine-level layer employs model predictive control (MPC) for high-precision power tracking. Furthermore, an intelligent task allocation algorithm based on multi-agent deep deterministic policy gradient (MADDPG), a distributed consensus based coordination mechanism, and an intelligent fault tolerance strategy including detection, isolation, and gradual recovery are proposed. Simulation results show that the method reduces regulation cost by 15%, controls power tracking error within 1.2%, completes rescheduling within 12 seconds, and maintains system availability above 80% under severe faults, significantly outperforming traditional proportional allocation.