Reactive-Voltage Coordinated Control of Offshore Wind Farm Based on Deep Reinforcement Learning

The capacitance effect of submarine cables increases the risk of voltage overrun of offshore wind farms. The coordinated control of reactive-voltage (Q-V) is an effective way to improve the voltage stability. The existing research focuses on the Q-V control method based on reactive optimal power flow (Q-OPF) theory. However, there are still two problems: the accuracy and speed of wind farm OPF model are difficult to guarantee. Based on this, a reactive power voltage control method for offshore wind farm based on deep reinforcement learning is proposed. Firstly, an optimal control model of reactive power flow is established to improve the voltage stability of wind farm while considering the system power loss. Then, the optimal control model of voltage is transformed into a Markov game process. Finally, the optimal control model is trained by using the deep deterministic policy gradient (DDPG), and the method does not need to rely on historical data. Simulation results show that the proposed method can effectively improve the voltage stability of wind farm, and has better model solving accuracy and speed performance than traditional methods.