Deep Reinforcement Learning-Based Voltage Control Method for Distribution Network with High Penetration of Renewable Energy

In order to improve the stable operation of the distribution network that is connected to renewable energy and solve the problem of the distribution network with voltage out of limit, an optimal control method of distributed generation participating in voltage regulation based on Deep Deterministic Policy Gradient (DDPG) algorithm is proposed in this paper. Firstly, the application basis of DDPG algorithm is presented, after that, a mathematical model of the environment of the distribution network with distributed generations based on DDPG algorithm is established. Finally, based on the principle of reactive power compensation as the mainstay and active power reduction as a supplement, the voltage regulation control process is given. The system can continuously accumulate experience through off-line learning, the optimal control is given according to the real-time measurement data of the distribution network. Meanwhile, the speedability and effectiveness of the approach proposed are verified by simulation.