Research And Implementation Of Network Security Situation Awareness System Based On Neural Network

With the popularization of Internet,mobile Internet and Internet of things,the current network environment is becoming more and more complicated.Therefore,the types and severity of network security threats are increasing,and the traditional means of network security protection are facing great challenges.However,network security situation awareness can not only timely monitor the security state changes of the current network environment,but also effectively predict the network development trend in the future based on the collected network historical data.Therefore,network security situation awareness technology has great research value.This paper conducts an in-depth study on situation assessment and situation prediction in network security situation awareness technology.The research contents of this paper are as follows:1.Analyze and combine relevant data contents of the real network defense data set(cse-cicids2018)to build the network security situation assessment indicator system.This index system divides the situation elements into four aspects: vulnerability,disaster tolerance,threat and stability,and puts forward the quantitative formula of nominal second-level indicators,so that the situation elements can comprehensively measure the security state of the network environment from different dimensions.2.An improved support vector machine(SVM)model parameter algorithm based on genetic algorithm is proposed to realize network security situation assessment.By analyzing the characteristics of linear kernel function and gaussian kernel function,a new fusion kernel function is realized,which can dynamically give consideration to both global and local features.At the same time,in order to solve the disadvantages of traditional SVM algorithm using grid search for parameter optimization,such as low efficiency and falling into the local optimal solution,this paper adopts genetic algorithm to optimize the penalty parameters C,times d,bandwidth ? and weight coefficient ? of the improved SVM algorithm.Experimental results show that the scheme can achieve 95% accuracy in small sample space.3.An algorithm based on Elman neural network parameters such as SAPSO algorithm optimization and improvement is proposed to realize network security situation prediction.In this paper,on the basis of traditional double hidden layer structure is improved,by increasing the sequence layer,the output results are within the part time before storage,random and time sequence weight accumulation,we pass to make history the output of hidden layer can have influence on model training conforms to the periodic characteristics of network security situation.At the same time,aiming at the shortcoming of neural network training that it is easy to fall into the local optimal solution,SAPSO algorithm is adopted to optimize the parameters of Elman neural network,so as to realize accurate prediction of network security situation.Through experimental analysis,the scheme has a good accuracy of situation prediction.