AI discovers better ways to fast-charge batteries
AI discovers better ways to fast-charge batteries
Charging batteries quickly without wearing them out is hard—and testing each new idea takes time and money. Researchers show that large language models (the tech behind chatbots) can help design smarter charging “recipes.”
- Two approaches: Prompt-to-Optimizer (P2O), where an AI writes small programs that learn a charging pattern, and Prompt-to-Protocol (P2P), where it directly writes the current profile with a few tunable numbers.
- Closed loop: The AI proposes, tests, and improves recipes without needing gradients or huge constraints on the search space.
- Results: In realistic fast-charging tests, both methods improved battery state-of-health by ~4.2% over a strong multi-step constant-current baseline, with P2P achieving this under the same number of protocol evaluations. P2O also beat Bayesian optimization, evolutionary algorithms, and random search.
Why it matters: AI can explore a wider range of charging strategies, honor plain‑language constraints, and make costly lab optimization more efficient.
Paper: From Prompt to Protocol: Fast Charging Batteries with Large Language Models (arXiv: https://arxiv.org/abs/2601.09626)
Paper: https://arxiv.org/abs/2601.09626v1
Register: https://www.AiFeta.com
AI Batteries FastCharging LLM MachineLearning EnergyStorage Research Optimization