Battery Energy Storage State-of-Charge Forecasting: Models, Optimization, and Accuracy

Abstract

Battery energy storage systems (BESS) are a critical technology for integrating high penetration renewable power on an intelligent electrical grid. As limited energy restricts the steady-state operational state-of-charge (SoC) of storage systems, SoC forecasting models are used to determine feasible charge and discharge schedules that supply grid services. Smart grid controllers use SoC forecasts to optimize BESS schedules to make grid operation more efficient and resilient. This paper presents three advances in BESS SoC forecasting. First, two forecasting models are reformulated to be conducive to parameter optimization. Second, a new method for selecting optimal parameter values based on operational data is presented. Last, a new framework for quantifying model accuracy is developed that enables a comparison between models, systems, and parameter selection methods. The accuracies achieved by both models, on two example battery systems, with each method of parameter selection are then compared in detail. The results of this analysis suggest variation in the suitability of these models for different battery types and applications. The proposed model formulations, optimization methods, and accuracy assessment framework can be used to improve the accuracy of SoC forecasts enabling better control over BESS charge/discharge schedules.