Inspection of rail transit safety protection zones is a critical measure for ensuring urban rail transit safety. However, in complex nocturnal environments, traditional manual inspections and UAV visible light technologies struggle to meet inspection requirements. To address these challenges, this paper proposes an intelligent monitoring method integrating thermal infrared imagery technology with the YOLOv5 deep learning algorithm. The approach involves equipping unmanned aerial vehicles (UAVs) with thermal infrared cameras for nocturnal image acquisition in protection zones. The YOLOv5 model is employed for target detection of construction equipment in thermal infrared imagery, while thermal anomaly feature extraction, dynamic background modeling, and a multi-rule classifier are comprehensively applied to achieve automatic identification and judgment of construction equipment status. This study selects the Liuchen Road Depot of Shanghai Metro Line 21 as the experimental area, conducting test research through route design, dataset preparation, model training, and performance evaluation. Experimental results demonstrate high accuracy in equipment identification and status monitoring, with the overall model algorithm precision reaching 94.33% and a Kappa coefficient of 0.89. This method effectively fulfills the demands of nocturnal rail transit inspections, enabling precise recognition of nighttime construction equipment in safety protection zones, early warning of abnormal environmental temperature changes, and discrimination of equipment operating states. It provides a feasible technical solution for implementing nocturnal inspections of urban rail transit systems.