The combination of deep learning and intrusion detection technology has brought better security protection to industrial control networks.Federated learning has become a hot research topic in the field of industrial control intrusion detection by employing data from multiple parties to train an efficient model under the premise of protecting user data privacy.Aiming at the problem of low detection rate of intrusion detection system due to high dimension,feature redundancy,and lack of attack samples of industrial control network traffic,an industrial control network intrusion detection system based on deep learning and federated learning is constructed.Firstly,a multi-model fusion intrusion detection model is proposed,use weight-tied stacked autoencoder and single layer convolutional neural network to initially extract network traffic features,and use residual convolutional neural network and residual bidirectional gated recurrent unit to further extract spatial features and temporal features of network traffic,the extracted two features are fused to identify network attacks.Secondly,an industrial control network intrusion detection system based on the above model and federated learning is established,and the effectiveness of the system is verified using water storage tank dataset and gas pipeline dataset.The experimental results show that the industrial control network intrusion detection system based on deep learning and federated learning can improve the detection rate of the intrusion detection model and provide better security protection to the industrial control network.
industrial control networkintrusion detectiondeep learningfederated learning
NETL
NSTL
万方数据
