An Intrusion Detection Scheme Combining FCM and Kohonen Network
Chen Ming-Xia, Zhang Han, Li Shun-Yan · 2019
In order to solve the shortcomings of traditional industrial control network intrusion detection schemes, such as insensitivity to detection samples and inaccurate judgment of internal anomalies. An industrial control network intrusion detection scheme based on FCM algorithm and supervised Kohonen is proposed. FCM algorithm, FCM-GRNN network algorithm, FCM-BP network algorithm, FCM-Kohonen network algorithm and FCM-S_ Kohonen network algorithm are built on MATLAB software platform to test DARPA data samples. The accuracy of clustering results of different types of intrusion is counted according to five algorithms. The scheme can detect NORMAL, U2R, R2L, DoS and PRB network attacks more accurately, and the overall average classification accuracy rate is more than 95%.