A review on AI-powered safety management of battery energy storage system

Abstract

The battery energy storage system (BESS) has become a key element for modern power infrastructure, renewable power plants, and artificial intelligence (AI) data centers. BESS assembles massive high-capacity Lithium-ion battery cells into MWh-scale container units and power stations, amplifying the risk of thermal runaway and fire hazards. This paper first reviews the BESS safety issues and incidents across different lifecycle stages, including fire events during manufacturing, transportation, operation, emergency response, and recycling, and analyzes these incident causes and statistics over the past three years. Then, we establish an AI-powered BESS safety management framework, covering the whole lifecycle from design and operation to emergency response and end-of-life utilization. The review highlights the broad applications of AI in the intrinsic safety material selection, internal aging mechanisms modeling, and manufacturing quality control. With the massive operational data, AI can also support fault diagnosis, system aging prognostic, battery thermal management, and thermal runaway warning. We also review intelligent strategies for fire/explosion mitigation and second-life battery safety management. Furthermore, key technical challenges and prospectives of future research are discussed, covering the lack of dedicated BESS data, practical deployment of novel large language models, and application of new digital twin technologies. This review provides a valuable framework for improving the intelligence in BESS safety management and battery-based powering system across their entire lifecycle.