An integrated multi-timescale MPC framework for coordinated wind farm power maximization, load management, and grid support

Modern wind farms face challenges maximizing energy capture amidst wakes, minimizing loads, ensuring grid integration (AGC), and handling uncertainty, often poorly addressed by conventional controls. This paper proposes a novel three-stage, multi-timescale Model Predictive Control (MPC) framework. Stage 1 employs wind predictions for large-timescale predictive yaw optimization, maximizing long-term energy while proactively reducing yaw actuation and considering fatigue. Stage 2 uses small-timescale stochastic MPC for real-time refinement. Stage 3 delivers efficient micro-timescale power tracking, minimizing both error and control action duration. This enhances the speed and accuracy of AGC response, thereby improving the wind farm’s contribution to grid stability and ancillary services. Supporting algorithms enhance computational efficiency. Case studies using real wind data and diverse layouts demonstrated significant power gains (2.3–5.2% over MPPT). Crucially, yaw actuation remained comparable to MPPT levels (e.g., −1% for 28 turbines) yet drastically reduced (up to 69%) versus reactive optimization, achieving a superior balance between energy gain and wear mitigation. Power tracking was accurate and significantly faster (e.g., 25 s less cumulative action time/hour vs baseline). This integrated framework offers a robust, efficient solution for advanced wind farm control, enhancing profitability and reliability.