Network traffic prediction method based on graph convolutional neural network

The invention discloses a flow prediction method based on a graph convolutional neural network, and belongs to the field of network security. The invention provides a flow prediction method based on a graph convolutional neural network. Network flow prediction is carried out by using a space-time graph convolutional network. According to the method, improvement is carried out on the basis of a traditional graph convolutional neural network, an attention mechanism is introduced, a full connection layer is replaced by a 1 * 1 convolutional layer, and meanwhile product operation is carried out on a time domain convolution result. The accuracy and efficiency of network traffic prediction are improved, so that the method has potential application prospects in the field of network traffic prediction. Experiments prove that the accuracy and the performance of the proposed model in network traffic prediction are superior to those of other neural network prediction models. Through the method, various information such