Intrusion Detection Method for Industrial Control System Based on Parallel CNN-LSTM Neural Network Improved by Self-Attention

As an important part of the information age, network security has been concerned. Industrial control system is an indispensable part of modern production process, but it is also facing the threat from cyber-attacks. Intrusion detection system is an indispensable part of the industrial control system security defense system. It detects the attack data by monitoring the real-time status of network traffic. This paper offers an intrusion detection approach based on parallel CNN-LSTM with self-attention mechanism in accordance with the features of industrial control system network traffic, using the UNSW-NB15 dataset as the research object. After experimental analysis, the accuracy, recall and F1 value are 98.66%, 95.88% and 95.91% and The FPR and FAR are 2.28% and 2.23%. Compared with other intrusion detection models, the proposed method has better performance.