A consistency-oriented adaptive power allocation method for battery energy storage systems under seasonal wind–photovoltaic scenarios

Abstract

Abstract With the continuous increase in wind and photovoltaic penetration, battery energy storage systems (BESSs) have become an important supporting resource for power balancing and flexible regulation in renewable energy stations. However, during the long-term operation of multi-cluster parallel battery systems, state-of-charge (SOC) inconsistency continuously accumulates owing to inter-cluster capacity differences, deviations in initial states, and operational nonuniformity. This accumulation further leads to unbalanced cluster responses, premature triggering of boundary constraints, and a reduction in the available regulation capability of the system. To address these issues, this study proposes a consistency-oriented adaptive optimal power allocation (COAO) method for seasonal wind–photovoltaic operating scenarios. First, daily operating feature vectors are constructed from annual wind power, photovoltaic, and load data, and fuzzy C-means clustering is employed to extract four typical operating scenarios corresponding to spring, summer, autumn, and winter, thereby establishing a unified cross-season evaluation framework. Second, for multi-cluster electrochemical energy storage systems, a consistency-oriented mathematical model is developed that incorporates cluster-level SOC dynamics, SOC safety boundaries, power constraints, ramp-rate constraints, and a directional exit mechanism. Furthermore, a cluster-level priority scoring function is constructed by comprehensively considering direction-related SOC margins and capacity differences, and adaptive weight adjustment together with feasible-region optimization correction is introduced based on consistency indicators. In this way, both power execution capability and consistency maintenance capability can be achieved under engineering constraints. Finally, under the four seasonal representative scenarios, the proposed method is systematically compared with three benchmark allocation strategies, namely state-of-charge proportional allocation (SCPA), maximum residual power-based ranked water-filling (MRWF), and model-predictive-control-based SOC balancing (MPC-SOC), over 30 paired simulations. The results show that COAO consistently achieves the lowest mean SOC range and SOC standard deviation and also provides the best overall power-tracking performance across all four seasonal scenarios. This study can provide methodological support for the refined operational control of multi-cluster energy storage systems in renewable energy stations.