A Network Traffic Prediction for Information Security in the Smart Power

With the rapid development of network information technology, the volume of network data is increasing exponentially, and the scope of network applications is expanding. Network security issues are receiving increasing attention from the academic community. The integration of data science and artificial intelligence has intensified global network security challenges, especially in the Internet of Things (IoT) for power systems. As more power-related devices are connected to networks, security threats are heightened. Network traffic prediction is crucial for maintaining the information security of power IoT through proactive security measures. This paper proposes a novel model that combines BiGRU and Transformer to enhance long-term temporal dependency modeling. The model integrates Graph Convolutional Networks (GCN) for spatial features, BiGRU for time dependencies, and transformer-based feature weight assessment, improving the accuracy of anomaly detection and network traffic prediction. The model’s predictive capabilities are further enhanced by utilizing multi-head self-attention and large-scale language models. We demonstrate the model’s outstanding potential on real-world datasets.