Discovery Learning accelerates battery design evaluation

2025-08-09

AI Summary

In battery development, prototyping and cycle-life testing are time-consuming and costly, while existing data-driven approaches rely on labeled target-design data, hindering pre-prototype prediction. This work introduces Discovery Learning (DL), a novel paradigm that— for the first time—integrates educational psychology-inspired cognitive mechanisms into battery science machine learning. DL synergistically combines active learning, physics-guided learning, and zero-shot learning to enable cross-format prediction of cycle life for unseen material combinations using only historical design data. Crucially, it requires no labeled data from the target battery system. Validated on 123 industrial-grade pouch cells, DL achieves a mean absolute percentage error of only 7.2% in cycle-life prediction. Compared to conventional methods, it reduces experimental time by 98% and energy consumption by 95%, substantially accelerating battery design iteration.

Technology Category

• Machine Learning: Life-Long and Continual Learning
• Constraint Satisfaction and Optimization: Constraint Learning and Acquisition
• Search and Optimization: Learning to Search

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Fast and reliable validation of novel designs in complex physical systems such as batteries is critical to accelerating technological innovation. However, battery research and development remain bottlenecked by the prohibitively high time and energy costs required to evaluate numerous new design candidates, particularly in battery prototyping and life testing. Despite recent progress in data-driven battery lifetime prediction, existing methods require labeled data of target designs to improve accuracy and cannot make reliable predictions until after prototyping, thus falling far short of the efficiency needed to enable rapid feedback for battery design. Here, we introduce Discovery Learning (DL), a scientific machine-learning paradigm that integrates active learning, physics-guided learning, and zero-shot learning into a human-like reasoning loop, drawing inspiration from learning theories in educational psychology. DL can learn from historical battery designs and actively reduce the need for prototyping, thus enabling rapid lifetime evaluation for unobserved material-design combinations without requiring additional data labeling. To test DL, we present 123 industrial-grade large-format lithium-ion pouch cells, spanning eight material-design combinations and diverse cycling protocols. Trained solely on public datasets of small-capacity cylindrical cells, DL achieves 7.2% test error in predicting the average cycle life under unknown device variability. This results in savings of 98% in time and 95% in energy compared to industrial practices. This work highlights the potential of uncovering insights from historical designs to inform and accelerate the development of next-generation battery technologies. DL represents a key advance toward efficient data-driven modeling and helps realize the promise of machine learning for accelerating scientific discovery and engineering innovation.

Problem

• Accelerates battery design evaluation with Discovery Learning
• Reduces need for costly battery prototyping and testing
• Predicts battery lifetime without requiring labeled data

Innovation

• Discovery Learning integrates active, physics-guided, zero-shot learning
• DL reduces prototyping needs with historical battery data
• DL achieves 7.2% test error, saves 98% time