Application Research Of Network Security Situation Prediction Based On LSTM Neural Network

Public safety issues have always been the most important issues in the public production environment.Constructing a social public safety monitoring system to prevent social public safety incidents can shorten the emergency response time.Although it is very important to deal with it after an accident,no matter what the accident is,the best way is to prevent it in advance to avoid it.More and more enterprises are introducing network monitoring equipment into the public production environment,using the data monitored by the network equipment to prejudge the safety of the public production environment,and actively adopting corresponding emergency response measures.In this process,we must provide a reliable operating environment for network detection equipment in the public production environment to prevent the attack of malicious programs from causing data monitoring equipment failures and affecting the accuracy of data monitoring.Network prevention methods are constantly innovating,and network threats and attack methods are also intensifying.The serious imbalance between the development of the two has made information system security seriously challenged.Therefore,this article combines the history of network attack data with the current network security situation to predict the security threat situation in the future.This is conducive to maintaining the health of the network environment and creating a more secure construction site.In this paper,the theoretical basis of network security prediction and the commonly used network security prediction methods are studied in detail.For the specific network environment,a network security situation prediction model based on optimized Long Short-Term Memory neural network is proposed.The main work is as follows:(1)The LSTM neural network algorithm used in this paper overcomes the fact that the Recurrent Neural Network cannot process long-term dependent data,introduces the LSTM neural network into the situation prediction field,and uses the stochastic gradient descent algorithm as the minimum sample training.The goal of loss is to establish a network security situation prediction model to predict the future network security situation.(2)Combining the characteristics of LSTM neural network,the experimental results of multiple predictions are analyzed,and the data of the predicted completion is optimized by matrix transformation.According to the characteristics of the existing sample data,the optimized LSTM neural network security situation prediction model is constructed and a good prediction effect is obtained.(3)By comparing the simulation experiment with other time series algorithms,the accuracy and performance of the network security situation prediction based on the optimizedLSTM neural network prediction model are verified.Simulation experiments show that compared with other commonly used time series methods,the prediction model proposed in this paper has higher prediction accuracy,which can reflect the overall situation of network security situation more intuitively and provide a new solution for network security situation prediction.