Core Viewpoint - The recent publication of the research paper "Discovery Learning predicts battery cycle life from minimal experiments" by Fulin Technology and a professor from the University of Michigan introduces a groundbreaking machine learning method that significantly reduces the time and energy required for battery life validation, addressing high development costs and long cycles in battery development [1][3][4]. Group 1: Discovery Learning Method - The "Discovery Learning" method compresses traditional battery life validation to 33 days and 0.468 MWh, saving 98% of time and 95% of energy, thus shortening the R&D cycle from years to months [3]. - Only 51% of the prototype's first 50 cycle data is needed to achieve results that exceed current mainstream methods, significantly reducing trial and error costs in R&D and manufacturing [3]. Group 2: Impact on Fulin Technology - The introduction of "Discovery Learning" not only represents a technological breakthrough but is also closely linked to the overall development of Fulin Technology, enhancing its competitiveness in the battery technology R&D field [4]. - Fulin Technology has established a flexible manufacturing platform capable of commercializing various battery types and materials, including high-nickel ternary, lithium iron phosphate, and sodium-ion batteries [3][4]. Group 3: Future Directions and Market Position - Solid-state batteries are viewed as the next core direction for power batteries, with Fulin Technology aiming to lead in this area through extensive R&D and industrialization efforts [5]. - The company has developed a diverse and high-quality customer ecosystem, including major clients such as GAC, Dongfeng, and a leading eVTOL company, which positions it well to expand market share and strengthen growth [5][6]. - Fulin Technology plans to continue its innovation-driven development approach, deepening collaborations with top research institutions and integrating cutting-edge technologies into its multi-material battery R&D and production processes [6].

美国密歇根大学工程学院科学家在近期出版的《自然》杂志发表论文，宣布开发出一款基于"发现学 习"理念的人工智能（AI）工具。这款全新的智能体仅需几天实验测试数据，即可准确预测新电池的循 环寿命。 团队使用美国Farasis能源公司的袋式电池数据对该模型进行了验证。尽管训练集仅包含类似AA电池的 圆柱形电池，系统仍预测出了结构更复杂、尺寸更大的袋式电池性能。这意味着该方法具备良好的泛化 能力，适用于多种电池形态。 团队表示，这项技术未来可拓展至电池安全、快充性能等更多维度。更重要的是，"发现学习"作为一种 新型机器学习范式，有望推广至化学、材料科学等高度依赖昂贵实验的领域，为那些长期受限于高成 本、长周期的研究按下"加速键"。 （文章来源：科技日报） 该系统的灵感源自一种"边做边学"的"发现学习"认知模式，其通过实践探索获取知识，而非被动接受理 论灌输。团队将其引入AI领域，打造出这款AI智能体。 具体而言，这套系统由3个核心模块协同工作。"学习器"负责提出问题，决定建造哪些电池原型，并进 行短周期测试以填补知识盲区；"解释器"分析历史数据，结合物理模型模拟电池内部反应，挖掘不同电 池间的共性规律；"智囊"则综合 ...