Reliable and safe operation of batteries is increasingly challenged by diverse operating conditions and stringent demands for system resilience.Artificial intelligence(AI)has emerged as a transformative enabler of bat...Reliable and safe operation of batteries is increasingly challenged by diverse operating conditions and stringent demands for system resilience.Artificial intelligence(AI)has emerged as a transformative enabler of battery health management,offering capabilities beyond traditional models.This review provides a structured synthesis of recent progress in AI-enabled diagnostics.Advances in state estimationincluding state of health(SOH)and remaining useful life(RUL)-are first examined,with methodological breakthroughs identified across diverse task formulations.The evolution of AI architectures is then traced,from conventional neural networks to attention-based Transformers,physics-informed models,and federated learning,with particular attention to emerging paradigms such as foundation models,neuro-symbolic reasoning,and quantum machine learning that promise improved robustness and interpretability.To bridge laboratory innovation with deployment,a domain-adaptive four-stage data pipeline has emerged as a promising framework for real-world BMS signals-spanning operational segmentation,multi-scale denoising,degradation-aware feature engineering,and structured sample construction-designed to enhance generalization under heterogeneous and noisy conditions.Looking forward,a technological roadmap is outlined that integrates edge AI,digital twins,AIOps,quantum computing,wireless sensing,and self-repair systems.Collectively,these innovations transform batteries from passive energy reservoirs into intelligent cyber-physical agents endowed with perception,autonomous decision-making,and resilient fault response-paving the way toward truly battery-centric autonomous energy systems.展开更多
基金funded by the Independent Innovation Projects of the Hubei Longzhong Laboratory(2022ZZ-24)。
文摘Reliable and safe operation of batteries is increasingly challenged by diverse operating conditions and stringent demands for system resilience.Artificial intelligence(AI)has emerged as a transformative enabler of battery health management,offering capabilities beyond traditional models.This review provides a structured synthesis of recent progress in AI-enabled diagnostics.Advances in state estimationincluding state of health(SOH)and remaining useful life(RUL)-are first examined,with methodological breakthroughs identified across diverse task formulations.The evolution of AI architectures is then traced,from conventional neural networks to attention-based Transformers,physics-informed models,and federated learning,with particular attention to emerging paradigms such as foundation models,neuro-symbolic reasoning,and quantum machine learning that promise improved robustness and interpretability.To bridge laboratory innovation with deployment,a domain-adaptive four-stage data pipeline has emerged as a promising framework for real-world BMS signals-spanning operational segmentation,multi-scale denoising,degradation-aware feature engineering,and structured sample construction-designed to enhance generalization under heterogeneous and noisy conditions.Looking forward,a technological roadmap is outlined that integrates edge AI,digital twins,AIOps,quantum computing,wireless sensing,and self-repair systems.Collectively,these innovations transform batteries from passive energy reservoirs into intelligent cyber-physical agents endowed with perception,autonomous decision-making,and resilient fault response-paving the way toward truly battery-centric autonomous energy systems.
