Cost-Effective Volt-VAR Control via Transactive Energy: A Data-Driven Approach

Abstract

To enhance voltage profiles in three-phase active distribution networks (ADNs), a data-driven Volt-VAR control (VVC) strategy within the transactive energy (TE) framework is proposed. In this framework, the distribution system operator (DSO) employs dynamic transactive prices to incentivize the third-entity-owned microgrids (TMGs) to participate in VVC. In the proposed framework, the neural networks (NNs) are firstly utilized to fit the historical transactive data to simulate the transactive behaviors of TMGs. Then, the transactive pricing strategy for TMGs is optimized through a multi-agent deep reinforcement learning (MADRL) algorithm. The application of NNs and MADRL efficiently addresses the critical privacy concerns of TMGs and the non-convexity of three-phase power flow. Finally, an index termed the levelized cost of VVC (LCOV) is proposed to verify the cost-effectiveness of the proposed TE-based VVC strategy. The effectiveness and advantages of this TE-based VVC strategy are validated on a modified three-phase IEEE 123-node system.