image: The framework positions an AI-driven large model as a unified intelligence hub, connecting materials, electrochemical systems, battery design, manufacturing/quality control, and operation across the battery value chain.
Credit: ©Science China Press
Recently, a cross-border collaborative team consisting of Sunwoda Mobility Energy Technology Co., Ltd (a globally leading battery manufacturer), Chery Automobile Co., Ltd (a world-renowned vehicle manufacturer), the State University of New York at Binghamton (including Professor M. Stanley Whittingham, a Nobel laureate), Semitronix Corporation (a globally renowned EDA company), the University of Delaware and Advance Power jointly officially published their review article titled "Revolutionizing Batteries Based on Digital Twin through AI-Simulation Synergy for Design, Manufacturing, Operation, and Recycle" in the international academic journal National Science Open.
The research team innovatively proposed the "Integrated Battery Large Model"， establishing the first AI-driven paradigm covering the entire lifecycle of the lithium-ion battery industry, providing a novel technological path for the industry's intelligent upgrade. The Battery Large Model system constructs a knowledge graph spanning the entire industrial chain by deeply integrating multi-source information such as material properties, cell design, manufacturing processes, and operational data. This knowledge base not only contains comprehensive data from the atomic scale to the system level but also incorporates the operational characteristics of batteries under different application scenarios, providing solid industry data support for the large model. Through continuous learning and knowledge updating, the system can continuously optimize the understanding of the complex relationships between battery materials, structure, and performance, laying a solid foundation for industrial innovation.
The system innovatively integrates time-series models, AI-physics hybrid models, multi-modal AI models, and reinforcement learning algorithms. This technical architecture enables the large model to simultaneously process diverse types of information such as text reports, microscopic images, sensor data, and simulation results, achieving deep perception and accurate prediction of the battery's full lifecycle status. Especially in key areas like cell design, lifespan prediction, and process optimization, the synergistic effect of multi-modal AI significantly enhances R&D efficiency and accuracy, bringing revolutionary changes to traditional battery R&D models.
The Battery Large Model is expected to achieve significant breakthroughs in autonomous design scheme generation, accurate performance prediction, and intelligent defect detection through AIGC technology. The system can automatically generate optimized electrode material formulas, cell structure designs, and manufacturing process parameters based on specific application scenario requirements, substantially shortening the R&D cycle. Simultaneously, the large model, trained on massive datasets, can accurately predict battery performance under different operating conditions and enable intelligent diagnosis of manufacturing defects and operational anomalies, significantly improving product quality, safety, and reliability.
The establishment of the Battery Large Model shall mark the official entry of the lithium battery industry into a new era of "pre-trained large models + professional domain knowledge." This innovative framework can bridge the critical path from basic research to industrial application through end-to-end intelligent solutions. Compared with traditional R&D models, the new system will not only greatly enhance R&D efficiency but also effectively reduce R&D costs and risks through data-driven intelligent decision-making. More importantly, the system can provide technical support for the refined management and optimization of the entire battery lifecycle, enabling collaborative innovation across the entire chain—from mineral processing and material development to product design, manufacturing, application, and recycling—thereby efficiently achieving resource recycling and fostering a sustainable ecosystem.