Research On Industrial Internet Security Situation Awareness Method Based On Machine Learning

The security issues of industrial Internet have been on the rise in recent years du e to the deep integration of industrial Internet and traditional networks.However,Traditional network defense measures are hard to comprehend in terms of the entire system’s operation.In comparison,network security situation awareness technology has a more comprehensive outlook,allowing for a more thorough analysis and assessment of threats,thus enhancing the monitoring capacity of the network.This thesis will explore the security situation awareness method of industrial Internet,based on research from traditional networks,by utilizing Support Vector Machine,Mayfly Algorithm,and Gated Recurrent Unit neural network methods.Its main contents are as follows:1.This thesis introduces the traditional network security situation awareness into the industrial Internet,and establishes an industrial Internet security situation awareness model based on SVM-GRU.To facilitate understanding of the model,a formal description of its implementation process is provided.Fisher score combined with kernel principal component analysis was also employed to eliminate redundant industrial Internet data,thereby reducing complexity and enhancing the computational efficiency of subsequent assessment methods.Experimental results demonstrate that this method reduces dimension of data features,saves computational cost,and enhances SVM’s recognition ability in comparison to other methods.2.Aiming at the problems that the existin g industrial Internet security situation assessment is inaccurate,time-consuming,and difficult to apply to the environment of industrial Internet,this thesis proposes an industrial Internet security situation assessment method based on improved mayfly algorithm optimized SVM.Tent chaotic map and mutation strategy were employe d to refine the basic mayfly algorithm,thus reducing its local convergence.Subs equently,the improved mayfly optimization algorithm was utilized to optimize the parameters of SVM,thereby finding the best SVM parameters and obtaining more precise classif ication results.Finally,the security situation of the present industrial Intern et was evaluated to gain a better comprehension of the security situation of the current network.Experimental results demonstrate that the proposed method is more precise tha n other methods and can more effectively fulfill the requirements of industrial I nternet security situation assessment.3.A proposal for an industrial Internet security situation prediction technique based on self-attention and GRU is made to tackle the difficulty in constructing current methods and the lack of accuracy.To begin,the self-attention mechanism combines industrial Internet time series data to create atten tion weights.Then,the data with attention weights is input into the GRU neural netwo rk to analyze the correlation between the time series data.Finally,the trained model is used to predict the security situation of the industrial Internet,and the final predicted security situation value is output.Experimental results show that compared with the existing methods,the proposed method has faster convergence speed and higher prediction accuracy,and has better prediction effect.