Research And Application Of Network Security Situation Prediction Method Based On Parallel Deep Forest

With the rapid development of the Internet and information technology,network security issues have become prominent.The awareness,assessment,and prediction of the network security situation have become a crucial research direction in the field of network security.The traditional network security situation presents factor assignment methods including natural assignment method and manual assignment method.There are problems with the huge amount of calculation and human factors,which will affect the situation assessment value and ultimately affect the situation prediction result.Traditional network security situation prediction methods usually require large-scale data training and are very sensitive to missing data.The prediction method lacks theoretical support and a reasoning process.The prediction efficiency is low and the convergence speed is slow,resulting in inaccurate situation prediction results and low prediction efficiency.Unable to accurately and timely provide forward-looking information for network managers,resulting in a network environment at risk.The visual presentation of various information on the network security situation is usually incomplete,and it is difficult to reflect the status of various information on network security in a timely and comprehensive manner.To solve the above-mentioned series of problems,the research content of this thesis mainly includes the following three parts:(1)This thesis proposes a method for dynamically adjusting network security situation rendered elements.This method is based on the fuzzy analytic hierarchy process optimized by the stochastic gradient descent algorithm to dynamically adjust the weights of the network security situation presentation elements.The detailed calculation steps of the method are introduced.The network security situation presentation element weight adjustment is one of the network security situation assessments.An important research branch.This method can dynamically adjust the weight of the network security situation presentation elements and improve the calculation efficiency,which can effectively provide a reference for the network security situation assessment.(2)This thesis proposes a network security situation prediction method based on parallel deep forests.Due to the error interference of traditional network security situation prediction methods,it is difficult to train,lack of data sensitivity,and low training efficiency.However,this method is easier to train,has good performance,high efficiency,and scalability,supports small-scale training data,has good theoretical analysis,can be deployed in parallel,and can replace the shortcomings of traditional prediction methods.This thesis also designs a method based on the Spark distributed computing engine to parallelize the algorithm,to achieve the goal of fast processing massive data and efficient network security situation prediction.(3)This thesis designs a simulation data visualization system for the network security situation.First,carry out simulation experiments on network data,and then use the simulation data to visually display various network information,network topology,and network security situation,and query different sizes,so that network users and network managers can control network security.Take necessary measures to deal with network security threats and provide a reference for the visual display of various information on the network security situation.