| With the continuous innovation and evolution of Internet technology,various network security issues emerge in an endless stream,and the importance of network security has become increasingly prominent.Compared with traditional network security defense technologies,network security situational awareness technology can proactively identify network threats,evaluate current network security conditions,and predict future network security trends.However,in the current large-scale network environment,the diversification and frequent occurrence of network attacks lead to network security data showing multiple characteristics,multi-source heterogeneity and strong randomness,and the existing network security situation assessment and prediction methods are difficult to effectively guarantee network security.Therefore,this thesis is committed to solving the shortcomings of existing situation assessment and prediction methods,and uses neural network technology to conduct in-depth research on network security situation assessment and prediction.The main research work is as follows:(1)Considering the lack of intelligent design of the network security situational awareness system model for the current large-scale network environment,a new network security situational awareness system model is proposed in combination with the neural network.The model adopts a hierarchical structure,which consists of data acquisition layer,data processing layer,feature extraction layer,intelligent evaluation layer,intelligent prediction layer and visualization layer from bottom to top.Among them,the intelligent assessment layer and the intelligent prediction layer are the core of the model,which are used to provide comprehensive and accurate situation assessment and prediction.Finally,the corresponding prototype system is implemented according to the proposed model,and its effectiveness is verified.(2)In view of the fact that the existing network security situation assessment data set has the characteristics of high dimensionality and feature redundancy,the common situation assessment method has the problems of slow convergence speed and easy to fall into local optimum,which leads to limited assessment accuracy.A network security situation assessment method based on Principal Component Analysis(PCA)and Probabilistic Neural Networks(PNN)is proposed.This method first reduces the feature dimension of the situation assessment data set through PCA;then combines the characteristics of fast convergence and high classification accuracy of PNN to construct a network security situation assessment model based on PCA-PNN.At the same time,use Seagull Optimization Algorithm(SOA)to optimize the hyperparameters of PNN,that is,the smoothing factor,to improve the evaluation performance of the model.It is verified by experiments that this method can improve the accuracy of situation assessment more effectively than other similar methods.(3)For the existing network security situation prediction method based on Gate Recurrent Unit(GRU),it is easy to lose important situation information during the transmission of security situation information,which leads to poor prediction accuracy.Based on the idea of residual network,a network security situation prediction method based on Convolutional Neural Network(CNN)and Residual Gate Recurrent Unit(Res GRU)is proposed.This method firstly improves the network structure of GRU,and builds a residual GRU module by introducing residual connections to alleviate the problem of losing important information of situational data.At the same time,in order to further improve the prediction performance of the model,the convolution operation of CNN is used to extract the important information of the situation data.Experimental results show that,compared with other similar methods,this method can significantly improve the accuracy of network security situation prediction. |
