Interpretable Predictive Rule-Based Energy Management for Residential Balcony PV Systems with Hybrid Storage

Abstract

The increasing adoption of residential balcony photovoltaic (PV) systems introduces new challenges for energy management due to fluctuating PV generation and household demand, energy storage integration, dynamic electricity prices, and limited or restricted feed-in remuneration. At the same time, the low-cost nature of balcony PV systems requires energy management solutions with low computational and implementation complexity. Existing approaches often involve trade-offs between performance, interpretability, and computational or deployment effort. Therefore, this work proposes a lightweight and interpretable predictive rule-based energy management system for residential balcony PV systems with hybrid electrical and thermal storage. The proposed approach is compared against passive operation, conventional rule-based control, and model predictive control (MPC). Simulation results based on real-world residential datasets show that the proposed controller increases PV self-consumption, reduces grid export and electricity costs, and achieves performance close to MPC while maintaining low computational complexity, transparency, and suitability for embedded real-time implementation.