Railway infrastructure safety detection technology based on CNN algorithm model

To address the issues of low efficiency, insufficient accuracy, and difficulty in real-time monitoring in traditional methods for detecting the condition of railway infrastructure, this paper proposed a portable railway infrastructure condition detection design scheme based on artificial intelligence and 8K video analysis. The scheme integrated convolutional neural network (CNN) deep learning models and image processing technology. By constructing a CNN algorithm model, the CNN model learned the image features of normal and abnormal bolts, enabling automatic identification of the abnormal state of bolts. It performed frame-by-frame analysis on the collected 8K video, achieving detection and analysis of the bolt condition. Taking high-speed railway bolt anomaly detection as an example, the experimental results showed that the AUC value was as high as 0.586, performing best among all test models in this study. It could accurately detect the abnormal state of bolts, providing a new technological approach for the safety monitoring of railway infrastructure and having significant practical value.