Integrating transformer and MSCNN dual-branch architecture research on intrusion detection in industrial control networks
In response to the existing intrusion detection methods for industrial control networks,there are problems such as insufficient extraction capabilities for multi spatial features and long-distance temporal features of industrial control traffic.A dual branch architecture of Transformer and MSCNN is proposed for intrusion detection in industrial control networks.This model utilizes multiple convolution kernels of different sizes in multi-scale convolution(MSCNN)to extract multiple spatial features from industrial control traffic,expanding the learning range of industrial control traffic features.At the same time,the introduction of Transformer enhances the model's ability to extract long-distance temporal features in industrial control flow,further improving the performance of the model.Conduct experiments using the UNSW-NB15 and NSL-KDD datasets.The results show that compared with other methods,this model can extract more comprehensive and effective features,and has good detection performance and generalization ability.
industrial control networkintrusion detectionspatial characteristicslong distance temporal featuresMSCNNtransformer
万方数据
维普
