TD3-based Voltage Regulation for Distribution Networks with PV and Energy Storage System

The access to high penetration photovoltaic (PV) significantly increase. A voltage regulation strategy is proposed to alleviate the voltage overrun problem in the distribution network. The voltage fluctuation is suppressed by controlling the energy storage system (ESS) and static var compensator (SVC). First, a voltage regulation model of the distribution network with PV and ESS is established; then, the voltage regulation of the distribution network is modeled as a Markov decision process to mitigate voltage fluctuations by finding the best output point of SVC and energy storage. This paper solves the voltage regulation strategy by the twin delayed deep deterministic policy gradient (TD3). The proposed strategy compensates for the shortcomings of the traditional mathematical modeling strategy, which relies on a physical model and complete information input. The algorithm validation indicates that the proposed strategy can inhibit voltage fluctuations effectively, and the validity of the proposed strategy is verified by comparing it with other deep reinforcement learning algorithms.