计算机科学
电池(电)
可视化
渲染(计算机图形)
人工智能
可解释性
大数据
鉴定(生物学)
表征(材料科学)
钥匙(锁)
数据可视化
组分(热力学)
机器学习
数据科学
深度学习
特征提取
数据挖掘
特征(语言学)
计算模型
人工神经网络
实验数据
作者
Kaile Niu,Yifei Yan,Lei Cai,Ke Du,Tianxiao Sun,Jinhao Meng
出处
期刊:Energy and AI
[Elsevier BV]
日期:2026-11-01
卷期号:: 100964-100964
标识
DOI:10.1016/j.egyai.2026.100964
摘要
Lithium-ion batteries (LIBs) exhibit hierarchical structures spanning multiple length scales, from electrode-level architectures to particle-scale microstructures and atomic-scale local environments. Their electrochemical behavior arises from coupled physical and chemical processes across these scales, rendering visualization and quantitative analysis indispensable for establishing structure-property relationships. However, the rapid development of advanced characterization techniques has led to an unprecedented growth in both the volume and complexity of experimental datasets, increasingly exceeding the capacity of conventional analysis approaches to extract physically meaningful insights. Artificial intelligence (AI), with its powerful data mining capabilities, is well placed to address the shortcomings of traditional methods. In this Review, we systematically summarize recent progress in AI-assisted battery imaging-based characterization across electrode, particle ensemble, single-particle and atomic scales. We highlight how computer vision (CV) and related machine learning (ML) methods enable image enhancement, feature extraction and quantitative analysis, thereby facilitating the identification of structure-performance correlations and degradation evolution pathways that remain difficult to capture using traditional tools. We further discuss key methodological components, including dataset construction, annotation, model training, validation and interpretation, and outline emerging directions towards reusable and collaborative artificial intelligence frameworks for battery characterization. Overall, AI is transitioning from a data-processing tool to a central framework for multiscale battery characterization and materials design. By enabling standardized datasets, improving cross-system generalization, and integrating physical constraints, AI is expected to drive battery research towards a more predictive, interpretable, and iterative paradigm.
科研通智能强力驱动
Strongly Powered by AbleSci AI