Design of Visual Detection Algorithm for Loose Screw Faults in Internal Structure of Energy Storage Lithium‐Ion Battery Packs

In recent years, with the rapid development of renewable energy in China, the installed capacity of new energy storage systems—particularly lithium‐ion batteries—has grown significantly. However, several major safety incidents at energy storage stations worldwide have raised serious concerns about the reliability and safety of these systems. Traditional Battery Management Systems (BMS) are primarily designed to monitor structural and physical defects within the battery, but they are incapable of detecting externally invisible faults such as loose screws—an issue that often serves as an early indicator of battery failure. To address this limitation, this paper proposes a visual detection algorithm for identifying loose screw faults inside lithium‐ion energy storage battery packs. The algorithm leverages deep learning and image processing techniques, employing the YOLOv8 object detection framework. Enhancements are made to both the backbone and neck of the network, and attention mechanisms are introduced to improve the recognition of small‐sized objects such as screws. Experimental results demonstrate that the algorithm achieves a detection accuracy of up to 95.6%, enabling automatic identification of screw loosening and providing early fault warnings. This method offers a promising solution for intelligent operation and maintenance of energy storage stations.