Research On Intrusion Detection Method Of Industrial Control System Based On Improved TWSVM

At present,the new generation of information technology and manufacturing technology represented by mobile Internet,cloud computing,big data,Internet of Things and artificial intelligence are accelerating the integration,driving the manufacturing industry to the direction of digitization,networking,intelligence and service.However,while accelerating the in-depth integration and development of informatization and industrialization,the new generation of information technology has also brought increasingly serious information security issues.Intrusion detection,as a technology that can monitor and protect the information security of industrial systems in real time,is favored by many researchers.Intrusion detection,as a technology that can monitor and protect the information security of industrial systems in real time,is favored by many researchers.The essence of research on intrusion detection methods is behavior classification.Twin Support Vector Machine(TWSVM)is based on support vector machine(SVM),which has the advantages of fast training speed and strong generalization performance.It can solve classification and regression problems well.This study builds on the TWSVM model based on particle swarm optimization(PSO)based on careful study of a large amount of literature and relevant theoretical knowledge.First,for the measurement units of industrial network data intrusion eigenvalues are different,the characteristics of the samples are different,Normalize the industrial network data;secondly,to solve the problems of redundancy and high data dimension of the industrial network data,use principal component analysis(PCA)to reduce the dimension and feature extraction of the industrial network data;then,for the single kernel function Insufficient performance,construct a mixed kernel function composed of Gauss kernel function and Sigmoid kernel function;in addition,use the binary tree method to construct a multi-class twin support vector machine for classifying data types.Since the entire algorithm contains multiple unknown parameters,this study proposes an improved particle swarm optimization algorithm(GAPSO)combined with genetic algorithm.First,for the problem of poor global optimization of the PSO algorithm,this paper applies the crossover and mutation ideas in the genetic algorithm to the PSO algorithm.Then,for the problem that the PSO algorithm is easily caught in local convergence,dynamic nonlinear inertial weights are used.Parameters and dynamic learning factors improve the optimization performance of the algorithm.Finally,the optimized parameters are substituted into the TWSVM model,and the industrial network data set released by Mississippi State University is used for experimental simulation.The simulation results show that the intrusion detection method proposed in this paper has better detection performance than the comparison method.