Research On Internet Security Situation Prediction Technology Based On Neural Network

With the rapid development of country's Internet industry,network security and informatization are gradually advancing.At the same time,many network security issues have been exposed.The timely detection and handling of security risks that may exist during network operation is important for promoting the development of my country's network information industry There's important meaning.If traditional network security defense ideas such as firewalls,intrusion detection,and data encryption are used to improve the security of networks and applications,only passive defenses can be made against network security incidents.In order to predict the occurrence of network security incidents,passive defense is transformed into "active" defense,so that the damage that the network system may bear is reduced to a low point.As a result,network security situational awareness technology continues to flourish.By extracting and evaluating multi-source heterogeneous data and environmental factors,this technology predicts the overall security status of the network in the short-term future,which is conducive to the next decision-making analysis and improves network management efficiency.The traditional hierarchical network security situation assessment model only uses a large amount of IDS alarm information to evaluate and calculate services,hosts,and network systems,ignoring the correlation between alarm elements,resulting in insufficient accuracy and objectiveness of the assessment results.To solve the above problems,an improved hierarchical network security situation assessment model is proposed.First,use the alarm matching process and the network environment information to measure the alarm success rate;then establish appropriate fuzzy rules for the alarm threat degree,success rate and cycle,and use fuzzy inference to realize the complex nonlinear mapping relationship between the three,and obtain a Comprehensive alarm value;finally,calculate the evaluation value of service level,host level and system level.The traditional hierarchical network security situation assessment model uses the alarm information evaluation calculation of the intrusion detection system,ignoring the correlation between the alarm elements,resulting in the evaluation results not being accurate and objective.To solve the above problems,improve the hierarchical situation assessment model and add a fuzzy layer as the basis.First,under the condition of fully obtaining network environment information,use the alarm matching method to measure the value of the alarm success rate;then analyze the impact of the network security situation from the three aspects of alarm threat degree,success rate and cycle,and obtain the relationship between them.Finally,from the service level,host level,and system level,the comprehensive alarm situation value is calculated separately.According to the simulation results,the situation assessment model proposed in this paper has achieved good results in eliminating underreporting and misreporting information,and the assessment results are significantly better than traditional assessment methods in terms of overall and accuracy.The IPSO-LSTM situation prediction model introduced in this article is compared with other prediction algorithms.The prediction accuracy of IPSO-LSTM has been improved,which verifies the superiority of the prediction algorithm in this article.