Machine learning in laser welding of electric vehicle batteries: Process control and quality assurance

Laser welding serves as a pivotal technology for ensuring the structural integrity and electrical connectivity of electric vehicle (EV) battery packs, which directly influences product safety and mass production consistency. However, the inherent non-linearity of process dynamics, compounded by the limitations of traditional open-loop control and lagging offline inspection, presents significant challenges to achieving zero defect manufacturing. Emerging machine learning technologies offer a transformative data-driven paradigm to address these bottlenecks by enabling real-time monitoring and intelligent decision-making. This paper presents a systematic review of machine learning applications in battery laser welding and categorizes them into two core domains, including intelligent process control and smart quality assurance. For process control, the review discusses the transition from passive monitoring to active adaptation through multimodal sensing, parameter optimization and reinforcement learning-based closed-loop adjustment. Regarding quality assurance, the paper analyzes strategies for real-time defect diagnosis and the quantitative prediction of key mechanical and electrical properties, and explores the prospective correlation between welding process signatures and battery lifecycle performance. Furthermore, the review critically examines current challenges, including industrial data scarcity, model generalization and system integration. Finally, it highlights future directions such as physics-informed neural networks and digital twin technologies to realize the vision of intelligent manufacturing of EV batteries.