Distributed collaborative model predictive control based on computationally feasible mesh wind farms

Abstract

With the increasing scale of wind farms and the complexity of their operating environments, meshing has been introduced as an effective means to accurately characterize the flow dynamics, energy transfer paths, and wake effects in wind farms. The continuous expansion of the wind farm scale has significantly increased the computational burden of the system. In this paper, a new distributed cooperative control strategy is proposed to efficiently respond to the active power control commands of the grid transmission system operator (TSO) and reduce the computational burden of the system. The strategy is based on the mesh wind farm model WFSim, the breadth-first tree search (BFS) algorithm, and distributed model predictive control considering the influence of neighboring turbines(DMPC-INT). The traditional centralized optimization problem is transformed into multiple parallel, mutually coordinated local optimization tasks, and the turbines in the cluster perform inter-neighborhood communication to ensure that the shared turbines converge to the global optimization. This not only promotes effective cooperation among turbines, but also effectively improves the overall computational efficiency of the system. Simulation results preliminarily verify the effectiveness of the control strategy, which can track the grid power reference signal in real time and accurately, providing strong Active Power Control (APC) support for stable grid operation and effectively reducing the computational cost of the wind farm.