Efficient anomaly intrusion detection using Transformer based GAN network
Junyao Feng, Chao-hong Wang, Hao Xue, Lijun Zhang · 2024
With the rapid development of power grid terminals technology and the continuous increase in threats of network attacks, network anomaly intrusion detection for power grid terminals has become more complicated and crucial. However, due to the large number of network protocol fields and diverse network behaviors, existing methods still face challenges. Therefore, this text proposed a network anomaly intrusion detection method based on TransGAN, which combines the ideas of generative adversarial networks and Transformer models, to learn the correlation between different types of network intrusion behaviors and capture the dependency relationship of the network data. Besides, XGBoost was utilized for feature extraction to reduce computational cost without compromising detection performance. The experimental results show that the proposed anomaly detection method achieved an accuracy of 84.64%on NSL-KDD, while reducing the computational cost of training and testing by 9% and 6% respectively. It makes a valuable contribution to the research and practice of network security in the field of power grid terminals.