|
With the development of science and technology, Unmanned Aerial Vehicle (UAV) technology has played a good role in modern logistics, communications, military, agriculture, disaster relief, and other fields. However, the lack of endurance capacity of drones restricts their development in various fields. Therefore, research on drone charging technology has become a key factor in facilitating continuous breakthroughs in drone technology. To improve the accuracy of automatic charging path planning for drones operating in multi-drone states in substations, this paper proposes a neural model for multi-drone full-area reconnaissance path planning based on a neural network coupled with a deep reinforcement learning algorithm. To test the accuracy of the flight path for automatic drone charging in multi-drone areas obtained through neural networks, greedy strategies, and experience replay designed in this paper, simulation experiments were conducted. The results showed that the automatic drone charging model for substations designed based on the neural network and deep reinforcement learning algorithm reduced the charging path length by 8.9% compared to traditional drone path planning algorithms. It lowered energy consumption during the drone's charging flight path, and the algorithm converged faster, demonstrating better path planning performance.
|
