State of power estimation for LIBs in electric vehicles: Recent progress, challenges, and prospects

Abstract

The rapid advancement of electric vehicle (EV) technology is revolutionizing the transportation, with electrification and intelligence serving as the primary driving forces. Accurate battery power estimation is crucial to this transformation. Lithium-ion batteries (LIBs), as the core energy storage components in EVs, exhibit strong nonlinear characteristics across multiple physical domains due to material properties and compatibility issues. As a result, accurate power estimation for LIBs poses a significant challenge in current EV development. This paper reviews state of power (SOP) estimation methods, categorizing them into four major types: characteristic maps, models, data-driven machine learning, and multi-state joint estimation. The principles, functionalities, and applications of each method are evaluated. This paper uncovers the underlying relationships among multiple states and elucidates why multi-state joint estimation outperforms single-state estimation. Furthermore, the fusion of physics-based models and data-driven models emerges as a promising direction for achieving high-precision SOP estimation under dynamic operating conditions. The challenges faced in SOP estimation are detailed, including the requirements for high accuracy, real-time performance, robustness, predictive capabilities, and safety margins. This study highlights four technical contradictions, such as balancing model complexity and real-time performance, and proposes a novel SOP estimation framework that leverages hybrid modeling and multi-state joint estimation. This new framework will bridge the gap between current estimation methods and the demands of intelligent EVs, thereby contributing to advancing the understanding of SOP estimation and ultimately enhancing battery performance, safety, and longevity.