Visible to the public Intrusion Detection Method of Industrial Control System Based on RIPCA-OCSVM

TitleIntrusion Detection Method of Industrial Control System Based on RIPCA-OCSVM
Publication TypeConference Paper
Year of Publication2019
AuthorsTong, Weiming, Liu, Bingbing, Li, Zhongwei, Jin, Xianji
Conference Name2019 3rd International Conference on Electronic Information Technology and Computer Engineering (EITCE)
Date PublishedOct. 2019
ISBN Number978-1-7281-3584-7
Keywordsanomaly detection, anomaly detection model, anomaly intrusion detection algorithm, feature extraction, ICs, industrial control, industrial control system, industrial control systems, industrial data sets, integrated circuits, Intrusion detection, learning (artificial intelligence), OCSVM, one-class support vector machine, outlier, particle swarm optimisation, pattern classification, principal component analysis, Protocols, pubcrawl, resilience, Resiliency, RIPCA, RIPCA algorithm, RIPCA-OCSVM, Robust Incremental Principal Component Analysis, Scalability, security of data, single classification problem, Support vector machines

In view of the problem that the intrusion detection method based on One-Class Support Vector Machine (OCSVM) could not detect the outliers within the industrial data, which results in the decision function deviating from the training sample, an anomaly intrusion detection algorithm based on Robust Incremental Principal Component Analysis (RIPCA) -OCSVM is proposed in this paper. The method uses RIPCA algorithm to remove outliers in industrial data sets and realize dimensionality reduction. In combination with the advantages of OCSVM on the single classification problem, an anomaly detection model is established, and the Improved Particle Swarm Optimization (IPSO) is used for model parameter optimization. The simulation results show that the method can efficiently and accurately identify attacks or abnormal behaviors while meeting the real-time requirements of the industrial control system (ICS).

Citation Keytong_intrusion_2019