AI-Driven Physics-Based Onboard Charger for Optimised Battery Management

Power-efficient battery management is critical for enhancing energy storage systems' performance, lifespan, and sustainability in modern technologies, including electric vehicles and renewable energy applications. This work introduces an AI-driven, physics-based onboard intelligent charger design that enhances battery charge and discharge processes and introduces assurance with its precision by leveraging physics data; the system predicts battery behaviour, adapts to varying conditions, and ensures optimal charging efficiency while reducing energy loss and preventing degradation. The solution incorporates a dynamic control system that adjusts charging parameters based on the state of charge (SoC), state of health (SoH), and thermal conditions of the battery, enabling adaptive and predictive control across diverse battery chemistries and operational requirements. The neural network (NN) training demonstrated stable convergence, achieving high precision in charge-discharge control through effectively optimising battery parameters. The system can also exhibit robust adaptability across diverse battery chemistries, including lithium-ion and lithium-iron-phosphate, and enhance the performance of second-life batteries.