Behavior Uncertainty Aware Operational Planning for Multi-Resource Virtual Power Plant in Balancing Market

Abstract

Aggregators in balancing markets operate Virtual Power Plants (VPPs) by integrating distributed energy resources like batteries, electric vehicles (EVs), and heat pump water heaters (HPs) installed in low-voltage households. A key challenge is managing user behavior uncertainty both in day-ahead bidding and real-time operation. This paper proposes a hierarchical planning framework using probabilistic scenario forecasting, two-stage stochastic programming, and model predictive control (MPC). Day-ahead bidding optimizes demand response (DR) capacity and resource schedules using two-stage stochastic programming under behavior uncertainty, while real-time MPC ensures commitment fulfillment. By explicitly linking scenario-based bidding with receding-horizon control, the framework increases bidding aggressiveness without compromising EV charging and hot-water availability. Evaluation with 30 households over 150 days shows the proposed method reduces net electricity cost by 14.1% and increases offered down-DR volume by 1.6 times compared to the robust method, while achieving a 98.69% DR fulfillment rate and limiting EV energy shortage and HP hot water shortage to 1.0% and 0.61%, respectively.