@article{Chen2026, 
author = {Tianxin Chen and Xin Lai and Fei Chen and Zhouyang Xu and Xuebing Han and Languang Lu and Yuejiu Zheng and Minggao Ouyang},
title = {Intelligent prediction of electrode characteristics based on neural networks in the lithium-ion battery production chain},
year = {2026},
journal = {Green Energy and Intelligent Transportation},
volume = {5},
number = {1},
keywords = {Lithium-ion batteries, Electrode manufacturing, Electrode characteristics prediction, Artificial intelligence, Permutation feature importance},
url = {https://www.sciopen.com/article/10.1016/j.geits.2025.100294},
doi = {10.1016/j.geits.2025.100294},
abstract = {Lithium-ion batteries (LIBs) are widespread with the fast development of new energy vehicles. The characteristics of LIB electrodes, including mass load, thickness, and porosity, are critical for battery performance such as energy density and lifespan. These characteristics are greatly influenced by the manufacturing methods and should be carefully considered during the production process development. However, the manufacturing process of electrodes is highly complex, involving a multitude of parameters. The traditional trial-and-error method has proven to be ineffective in improving manufacturing efficiency. In this study, we propose an artificial intelligence-based prediction method for estimating the key characteristics of electrodes. Specifically, it utilizes active material mass content, viscosity, solid-to-liquid ratio, and comma gap as input parameters. Compared to the traditional multiple linear regression method, the proposed method exhibits a significant improvement in accuracy. In certain cases, the root-mean-square error is reduced by an average of 35.5%, highlighting the superior prediction accuracy achieved by our method. Furthermore, we conduct a comparative analysis of different deep neural networks in predicting electrode characteristics. Finally, the importance of input features using the permutation feature importance analysis method is analyzed. By harnessing the powerful generalization ability of artificial intelligence, our method can be effectively applied to the manufacturing process of LIBs, resulting in a significant enhancement of battery production efficiency.}
}