| Industrial Internet has become a critical infrastructure in today’s era.While it promotes the digitization,intelligence,and networking of industry,the security problems it faces are becoming more and more prominent.In this thesis,based on the analysis of industrial Internet security,an early warning model of industrial Internet security situation based on a statistical information grid is established,and posture extraction and posture early warning is introduced into this model to deeply study the industrial Internet security protection methods,which can effectively avoid the security risk of the industrial Internet.The specific research contents are as follows:1.For the existing industrial Internet security situation early warning model when affected by virus attacks,network attacks,and other means,resulting in poor early warning effect and poor model responsiveness,this thesis establishes a piece of statistical information grid-based industrial Internet security situation,early warning model,by introducing a statistical information grid,which not only solves the problem of matching lags behind the attack process,but also makes the false alarm rate reduced,enhances the responsiveness of this early warning model,and is more conducive to the timely and accurate release of early warning.2.To overcome the limitations of traditional situational extraction methods for processing multi-featured and high-dimensional nonlinear data,this thesis proposes an industrial Internet security situational analysis element extraction method based on noise reduction self-encoder and improved random forest,to obtain more accurate and reliable results.Firstly,the noise reduction self-encoder is used for feature extraction,then the base classifier in the random forest algorithm is filtered,then the original random forest algorithm is improved by using weighted majority voting,and the reduced dimensional data is trained for classification to obtain the final classification results.The problem of imprecise and inefficient extraction of traditional situational analysis elements is effectively solved.3.To solve the problem that the existing situational warning techniques cannot accurately detect the network security situation and cannot detect the potential threats of the industrial Internet in time.This thesis first established a hidden Markov model based on compound attacks and then predict the next attack of intruders based on this model combined with techniques such as compound attack discrimination and attack intent identification.This early warning technique can effectively detect and warn of the change in the Industrial Internet security situation.The experimental results show that the situational analysis element extraction method and situational warning technology in the situational and warning model designed in this thesis can accurately classify different network attacks and predict the number and characteristics of attacks before they occur,to buy time for the security protection of industrial Internet systems and make effective responses to intrusions and attacks on time.The research results have theoretical and practical significance and promote the development of Industrial Internet security situational awareness technology. |
