Energy efficiency-oriented multi-agent deep reinforcement learning optimization method for unmanned aerial vehicle group

Abstract

The invention discloses an energy efficiency-oriented multi-agent deep reinforcement learning optimization method for an unmanned aerial vehicle group. The method comprises the steps of adopting an improved DQN deep reinforcement learning method based on Q learning, training and updating the neural network of each intelligent agent by using historical information of the unmanned aerial vehicle cluster to obtain channel selection and power selection decisions of each intelligent agent of the unmanned aerial vehicle cluster, training the neural network by using a short-time experience playback mechanism in the training process, and maximizing the energy efficiency value of the corresponding intelligent agent by using the optimization target of each neural network. According to the invention,a distributed multi-agent deep strong chemical method is adopted, and a short-time experience playback mechanism is set to train a neural network to mine a change rule contained in a dynamic networkenvironment. The problem that a convergence solution cannot be obtained in a large state space faced by traditional reinforcement learning is solved. Multi-agent distributed cooperative learning is achieved, the energy efficiency of unmanned aerial vehicle cluster communication is improved, the life cycle of an unmanned aerial vehicle cluster is prolonged, and the dynamic adaptive capacity of an unmanned aerial vehicle cluster communication network is enhanced.

Description

technical field [0001] The invention relates to the technical field of UAV cluster communication network access, in particular to an energy efficiency-oriented multi-agent deep reinforcement learning optimization method for UAV swarms. Background technique [0002] At present, the rapid development and application promotion of UAV technology is one of the frontier and hot issues, which has attracted extensive attention. Among them, the research on UAV swarms is the most eye-catching. UAV swarms can use low-cost UAVs to form groups according to different roles, and play a huge role in coordinated actions. [0003] But the key to the synergy of drone swarms lies in their robust communication network. Without a communication system to support the internal members of the UAV cluster, its coordinated action is out of the question. [0004] At the same time, the optimization of energy consumption of small UAVs, especially battery-powered UAVs, is crucial. The construction and ...

AI Technical Summary

Problems solved by technology

[0005] However, due to its dynamics, the energy efficiency optimization decision-making of UAV swarm communication faces severe challenges in large decision space.

Using traditional reinforcement learning methods, you will encounter the problem that the algorithm cannot converge or the convergence speed is too slow due to the large decision space.

Images