Fault Identification of Railway Signaling Lightning Protection Modules Based on Improved YOLOv8

Given the inherent inefficiency and limited coverage of traditional manual inspection methods for railway signaling lightning protection module fault detection, which cannot meet modern railway operation and maintenance efficiency requirements, this paper proposes an improved YOLOv8-based fault detection framework. By integrating MobileNetV4's lightweight architecture with the Convolutional Block Attention Module, the optimized model demonstrates improved detection accuracy and computational efficiency in complex scenarios. Experimental results demonstrate significant improvements over the original model in recall rate, precision, and mAP metrics. Specifically, GFLOPs are reduced from 28.7 to 8.2, while mAP@0.5:0.95 increases from 0.773 to 0.808, maintaining effective detection performance. This research provides a reliable technical solution for precise fault identification in railway signal protection systems, effectively ensuring equipment operational stability with substantial practical application value.