Research on Network Traffic Forecasting Method Based on AGDN DSTAM-LSTM Spatiotemporal Fusion Neural Network

With the rapid development of enterprise private networks, the key to enterprise network management is to ensure the stability and security of the network. Therefore, the research on using network traffic prediction technology to improve network security and stability has gradually attracted attention and has become a hot topic in the industry. At present, the network traffic prediction model based on deep learning still has room for improvement in accuracy, and it is necessary to make full use of the real network topology structure and enhance the extraction of information about time series characteristics from the actual traffic matrix data. These two deficiencies lead to the inaccuracy of network traffic prediction results, and the overall applicability and prediction of the model need to be improved urgently. Based on the background of traffic prediction in a dedicated network in a large-scale clean energy enterprise, an AGDN DSTAM- LSTM (Adaptive Graph Diffusion Networks and Long Short-Term Memory of Dynamic Spatio Temporal Attention Mechanism) prediction model based on the combination of adaptive graph diffusion network and dynamic spatiotemporal attention mechanism and long-term memory network is proposed based on the spatial topology and time series of traffic. The spatial and temporal characteristics of network traffic are captured, and the LSTM unit is used to integrate the historical state into the current cell state directly, and the ability of memory of historical flow is enhanced, and the network traffic can be fully predicted. At last, a verification analysis is conducted on the Abilene network traffic dataset to achieve accurate and effective prediction. The test results show that the model proposed in this study can effectively improve the prediction effect. This research is expected to provide reliable and efficient network traffic prediction for the private network of large clean energy enterprises and provide useful inspiration for the digital transformation of enterprises.