SCN security situation assessment and prediction model based on improved selective convolutional network

Abstract With the advancement of the Internet, cyber attacks are becoming increasingly opaque and complex. The traditional security defense system is unable to handle the rapid changes and updates of cyber threats. This study proposes a secondary classification network security situation assessment and prediction model based on an improved Selective Convolutional Network. The proposed model introduces the Convolutional Block Attention Module into the Selective Convolutional Network to enhance network environment perception and suppress redundant information. The model combines Gated Recurrent Unit and Naive Bayes to process and predict long sequence data. The model maintains temporal memory and improves the robustness and accuracy of prediction. In the situation assessment experiment, the recognition accuracy is 97.38%. The classification accuracy is 96.25%. The recall rate is 97.88%. The average task processing time is 0.54 ms. In the situation prediction experiment, the model achieves a mean absolute error of 2.63% in point prediction accuracy. In strong noise interference, the prediction accuracy is 90.25%. In severe attack behavior, the prediction accuracy is above 90%. The proposed model shows high assessment accuracy, prediction precision, and robustness. The results prove its effectiveness and advancement in the field of network security situation assessment and prediction. The model effectively addresses the lag problem of traditional methods in temporal prediction.