A Real-time Reconfiguration Approach for Wireless Energy Networks Using Heterogeneous Graph Neural Network

In recent years, drones have found widespread applications in communications, disaster relief, and other fields. However, due to battery capacity constraints, the limited flight endurance of drones hinders their full potential. With the development of wireless charging technology, it can effectively expand the operational range and duration of drones in various scenarios. This paper investigates wireless energy networks based on wireless charging, optimizing the wireless energy network comprising power generators, relays, and energy consumption endpoints to ensure reliable power supply for drones. To enhance system response times in disaster relief and other scenarios, we propose a real-time reconfiguration approach for wireless energy networks using heterogeneous graph neural networks. By modeling the wireless energy network through a heterogeneous graph neural network and utilizing reinforcement learning methods for model training, the trained model can achieve real-time construction of wireless energy networks. Comparative experiments demonstrate the effectiveness and efficiency of the proposed algorithm, which can facilitate real-time construction of wireless energy networks.