A Secure-Sustainable-Fast Charging Strategy for Lithium-Ion Batteries Based on a Random Forest-Enhanced Electro-Thermal-Degradation Model

In this article, a secure, sustainable, and fast charging strategy based on an advanced electro-thermal-degradation (ETD) model is proposed for lithium-ion batteries. In this ETD model, temperature, state-of-charge, and state-of-health are strongly coupled. It combines the simplicity of the equivalent circuit model and the high accuracy of the random forest (RF) algorithm. The RF algorithm is used to address uncertainties, such as measurement errors, parameter drifts, and parasitic effects, all of which may otherwise compromise model accuracy. The degradation model is derived from the Arrhenius equation and enhanced through data fitting techniques. To minimize charging time while respecting degradation constraints, we employ a data-driven Bayesian optimization (BO) approach with a constraint-based expectation improvement acquisition function. The proposed control strategy is implemented in a dual-loop structure, with an outerloop temperature control to ensure thermal safety and an innerloop current control to achieve sustainable and fast charging based on the BO. Experimental results demonstrate the effectiveness of the proposed strategy, showing that the ETD model achieves accurate temperature estimation across a wide range of operating conditions.