Nonlinear observation of battery microscopic states utilizing adaptive super-twisting sliding-mode observers based on a compact electrochemical-thermal-aging model

Abstract

Accurate microscopic-state estimation of lithium-ion batteries (LIBs) is a potential candidate for advanced battery management systems. This paper proposes an adaptive super-twisting sliding-mode observer (SMO) for LIB microscopic state perception using a compact and observable electrochemical-thermal-aging model. Considering the physical structure and mathematical model, a simplified single-particle mode with electrolyte is standardized as a compact state-space form after observability analysis. Results show that the compact model is capable of capturing the LIB terminal voltage and temperature with maximum errors of 39.8 mV and 0.2 °C, respectively. After that, the super-twisting SMO is integrated with a neural network-based adaptive law for addressing external disturbances and chattering. Estimation results show that during the majority of cycles, the errors of state of charge, volume-averaged concentration, temperature, and capacity are <3 %, 0.87 kmol/m3, 0.2 °C, and 0.1 Ah, respectively. The side-reaction overpotential of lithium plating prevention can be predicted. Comparable experiments with different observers and hardware-in-the-loop experiments are conducted to explore the estimation accuracy and real-time feasibility of the adaptive observer.