Research on Dynamic Security Detection of New Energy Terminals Based on Graph Neural Networks

With the rapid development of new technologies such as the power Internet of Things, as well as secondary businesses such as spot markets and source network load storage collaborative control, network boundaries continue to extend, node equipment greatly increases, functional applications become more in-depth, and data exchange becomes more extensive, resulting in a significant increase in the exposure of the power Internet of Things and higher requirements for grid security protection. Distributed new energy is widely connected to the grid, and massive heterogeneous load terminals are continuously connected. The traditional methods of relying on manual vulnerability mining, asset investigation, attack detection, and hidden danger disposal are difficult to apply, and there is an urgent need to form an efficient, easy to promote, and dynamic new model of network security protection. This article proposes an innovative two-stage anomaly detection method based on graph neural networks. In order to cope with the characteristics of flexible, user friendly, open interaction, and safe and controllable requirements of new energy power stations, technical research is carried out at three levels: dynamic security access on the power side, dynamic defense of centralized stations, and dynamic optimization of security facilities. The method proposed in this article achieves dynamic access control for various terminals, adaptive network transformation, and dynamic optimization of security facility strategies, supporting the improvement of network security dynamic protection capabilities