Multi-objective hierarchical co-optimization of battery capacity configuration and operational strategy for photovoltaic-battery systems in buildings

Proper configuration of battery capacity is critical for achieving desirable performance for distributed building photovoltaic-battery (PVB) systems. However, conventional battery capacity configurations are usually performed based on predefined rule-based operational strategies, of which the objectives are generally inconsistent with those of capacity optimization. To address this issue, this paper proposed a multi-objective hierarchical co-optimization (MHCO) framework for battery capacity configuration and operational strategy, which balances economic, technological, and environmental considerations. Within the proposed framework, the battery capacity configuration is positioned in the upper layer and the operational strategy in the lower layer, with both layers sharing a unified objective function. During each optimization step, the upper layer transmits updated battery capacities to the lower layer, which provides feedback on charge and discharge control sets and performance indicators. The optimization problem was solved using the second-generation Non-dominated Sorting Genetic Algorithm (NSGA-II) coupled with Dynamic Programming (DP). The proposed MHCO was comprehensively compared to the conventional method in terms of the optimized battery capacity (E_b,o) and the normalized gain (NG) of performance indicators. Based on the case study of a medium-sized apartment building, the MHCO achieved larger E_b,o, and the NG was up to 0.27, proving the necessity of aligning the objectives of capacity configuration and operational strategy. Energy flow analysis showed that the MHCO method has greater adaptability to various optimization requirements by unifying the objectives of battery capacity configuration and operational strategy. As a result, the MHCO method showed better performance under conditions of different PV penetration rates, dynamic FiT, and dynamic GEF, achieving maximum NG of 0.92, 0.31, and 0.80, respectively.