Stochastic Gradient Pulse Adaptation for Grid Friendly DC Fast Charging of Battery Electric Vehicles

This paper presents Stochastic Gradient Pulse Adaptation (SGPA), a real-time adaptive pulse-charging system for rechargeable electrochemical batteries that dynamically adjusts charging aggressiveness based on the battery's internal response, as opposed to predetermined CC–CV or fixed pulse profiles. SGPA is different from traditional charging methods that use static current de-rating and conservative voltage limits. Instead, SGPA uses gradient-based feedback from terminal voltage behaviour, temperature changes, internal resistance changes, and state of charge to continuously adapt pulse amplitude and duty cycle. This algorithm boosts the charging intensity when the electrochemical circumstances are good. It lowers the pulses slowly when signs of thermal or impedance-related stress show up. Simulation-based proof-of-concept experiments on a heavy-duty multi-battery system show that charging time is less than with multi-CCCV charging, while still keeping the current distribution across packs balanced. The suggested SGPA method adds an adaptive charging algorithm that is easy to understand and ready to use. It makes fast charging more efficient without lowering voltage and thermal safety limits.