基于深度学习的网络安全态势预测方法研究

Situational awareness has been a popular direction in the field of network security in recent years, situation prediction is based on the current situation assessment results and the existing historical assessment results data to predict the changing trend of the network security situation in the future, which is the ultimate goal of situation awareness.Aiming at the problems of the existing situation prediction algorithm, such as insufficient feature extraction ability, high sample requirements, and little attention to the influence of other features on the situation prediction results, a network security situation prediction method based on an improved long short-term memory network model is proposed.First, the input method of time series data is changed in the input layer of the model, and the influence of time characteristic factors on situation prediction is added to the input sequence, and then the self-attention layer is added after the hidden layer of the long-term short-term memory network to sssign weights to different characteristic factors, which improves the accuracy and stability of network security situation prediction.Through experiments, it was found that after changing the data structure of the input sequence, the accuracy of the model increased significantly; then the prediction effect of this model was compared with other deep learning models, which fully verified the accuracy and feasibility of using this model for situation prediction.