基于门控循环单元的局域网络总线入侵智能检测研究OA

In order to improve the timeliness and accuracy of laboratory LAN bus intrusion detection,a method of bus intrusion intelligent detection based on gate recurrent unit(GRU)is designed.The binary processing on qualitative features that only contain two states is conducted.The features containing three or more categories are converted into vector features by means of one-hot encoding.The standardized adjustments to the dataset are conducted to balance data features with different scales.In order to improve the detection limit,an undersampling algorithm combined with clustering is used to construct a balanced dataset.The GRU and convolution neural network(CNN)are integrated to construct CNN-GRU intrusion detection model,so as to achieve intelligent and efficient detection of local area network bus intrusions.The experimental testing results show that when detecting different attacks,both the Micro-F1 and Macro-F1 indicators of the designed method are relatively high,and the detection time for different attacks is less than 0.2 s.