The Method for Detecting Electrode Endpoints of Lithium-ion Batteries Based on Deep Learning

In the industrial production of high-energy-density lithium batteries, accurately detecting electrode endpoint positions is crucial for maintaining appropriate electrode spacing, as improper spacing can lead to short circuits or thermal runaway, causing significant safety issues. Due to the vague characteristics of electrode endpoints, traditional detection methods face challenges in achieving high precision. This paper proposes a deep learning-based method for electrode endpoint detection called the Lithium Electrode Detection Network (LEDNet). LEDNet utilizes the CSPDarkNet53 network as the backbone for feature extraction and employs a Path Aggregation Network (PAN) in the Neck layer for multi-scale feature fusion, effectively enhancing the perception of both local details and global information of electrode endpoints. Additionally, a novel loss function named PedLoss is designed to refine error adjustments for fuzzy electrode endpoints, thereby improving the detection capabilities for ambiguous endpoints. Experiments conducted on a self-constructed dataset demonstrate that the proposed method achieves an AP50 of 96% for the electrode endpoint detection task, significantly enhancing the accuracy of endpoint detection and thereby improving the safety and reliability of the lithium-ion battery production process.