Abstract: To improve the intrusion detection capability of railway networks and solve the problems of high false alarm rates, high maintenance costs, and inability to cope with unknown attacks in traditional network intrusion detection algorithms, this paper proposed a network intrusion detection model based on ResNet Att. This model combined the advantages of skip connections in ResNet networks with attention mechanisms to enhance its ability to identify abnormal behavior in network traffic. Through training and testing on the CICIDS-2017 dataset, the results show that the model achieves an accuracy of 99.75%, an average recall rate of 95.33%, and an average accuracy rate of 94.48% in multi classification tasks for network intrusion detection, all evaluation indicators surpass traditional network intrusion detection models. This helps to improve the network security of railway systems and provides technical support for the development of network security technology.