Distribution Network Reconfiguration Using Deep Reinforcement Learning

This paper proposes a deep reinforcement learning (DRL)-based framework for distribution network reconfiguration (DNR). The objective of the proposed framework is to minimize power losses in the network and various reliability indices including System Average Interruption Frequency Index (SAIFI), System Average Interruption Duration Index (SAIDI), and Average Curtailed Power (ACP). Constraints of the optimization problem are radial topology constraint and all nodes traversing constraint. The distribution network is modeled as a graph and the optimal network configuration is determined by searching for an optimal spanning tree. Contrary to existing analytical and population-based approaches, where the entire analysis and computation is to be repeated to find the optimal network configuration for each system operating state, DRL-based DNR, if properly trained, can determine optimal or near-optimal configuration quickly even with changes in system states. The Q-learning, a model-free reinforcement learning algorithm, is used by the proposed DRL-based framework to learn the action-value function. The effectiveness and efficacy of the proposed framework for DNR is demonstrated through a case study performed on 33-node distribution test system.