Data-Driven Modeling of Battery-Based Energy Storage Systems

Abstract

This article presents a data-driven modeling methodology applied to a battery-based power system comprising a power converter and an electric machine. The proposed method captures the dynamics describing the complete system and allows the identification of its parameters without the need for any explicit theoretical model of the components. In particular, the proposed approach considers the battery as the supplying element of a broader system comprising power electronics converters and direct-current motors, paying special attention to the battery open-circuit voltage curve estimation. This approach successfully yields a state-space representation that optimally describes the more essential variables, such as motor speed and output voltages of the converter and battery. Consequently, the proposed approach allows the generation of higher-order models representing transient and rapid dynamics and facilitates the identification of parameters that define reduced-order models describing slower dynamics. This streamlines the implementation of adaptive control strategies, providing an effective tool for their development and execution.