Multi-Stage Volt/VAR Support in Distribution Grids: Risk-Aware Scheduling With Real-Time Reinforcement Learning Control

The ever-increasing penetration of intermittent renewable resources in low-voltage power grids necessitates efficient operational strategies for voltage regulation as well scheduling the available resources. In this paper, a risk-aware Volt/VAR support framework followed by real-time reinforcement learning controller is presented three-phase distribution systems. stochastic stage, legacy regulating assets along with inverter-based photovoltaics (PVs) and energy storage system (ESS) are optimized considering day-ahead intra-day markets. Moreover, demand response (DR) reduction plans included proposed framework. By incorporating voltage-dependent load modeling study, implementation plan reduces consumption feeders running network at lower permissible limits. shiftable loads under DR program employed peak shaving to reduce costs. result shows that also dependencies on operation traditional devices. stochasticity abrupt changes PV generations represented Gaussian Mixture Model (GMM), indicating non-unimodal probability forecasting errors. scenario sets uncertain variables then reduced using fuzzy clustering technique. Decisions made scheduling, associated inverters ESS operation, revised controller, i.e., Deep Deterministic Policy Gradient (DDPG) learning. DDPG adopted control stage detailed unbalanced minimize deviation ramping ESS. performance multi-stage scheme verified active grid different scenarios.