Enhancement of Charging Speed and Analysis of Battery Configuration for Electric Vehicles Using Deep Learning

The initial models for lithium-ion battery inner states, consisting of coupled nonlinear partial differential equations (PDEs), encapsulate the system’s essential physics. Despite their accuracy in predictions, the computational demands of these models hinder their practical application for online control and monitoring. A refined model, capable of executing a full simulation within milliseconds, holds the potential to forecast internal battery states accurately. This study illustrates that the enhanced prototype is well-suited for real-time control applications. The modified form is employed to derive optimal operating rules (MPC) for the battery, demonstrating that the computational cost of each controller calculation is remarkably low, ensuring completion within the sampling time for analytical control. The prototype is further utilized to elucidate its application in state estimation MPC through observability experiments. Leveraging the moving horizon estimator (MHE) approach, which directly addresses limits and nonlinear dynamics, the reformulated model facilitates realistic and efficient calculations. The integration of this refined model with the battery management system (BMS) enables the construction and implementation of well-defined monitoring and control strategies, showcasing its practical utility in addressing real-world challenges.