飞控与探测2025，Vol.8Issue(1)：25-31,7.DOI:10.20249/j.cnki.2096-5974.2025.01.003
基于深度强化学习的飞行器过载和姿态智能控制研究
Intelligent Control of Aircraft Overload and Attitude Based on Deep Reinforcement Learning
摘要
Abstract
This paper addresses the problem of intelligent control of aircraft overload and attitude in complex and changing environments.It proposes a distributed intelligent agent control method based on the Soft Actor-Critic(SAC)algorithm,establishes a framework of distributed efficient environment interaction for deep reinforcement learning algorithms,and designs an intelligent con-trol algorithm system for aircraft overloadand attitude.This approach increases the scale and dis-tribution of training data for reinforcement learning algorithms,thereby improving the performance and robustness of aircraft control algorithms.Experimental results in simulated envi-ronments demonstrate that the trained intelligent agents effectively control overloadsandattitudes in unmanned aerial vehicle simulations.The distributed SAC algorithm outperforms the original SAC algorithm in controlling unmanned aerial vehicles in simulation scenarios.关键词
深度强化学习/无人飞行器/分布式SAC算法/过载控制/姿态控制Key words
deep reinforcement learning/unmanned aerial vehicles/distributed SAC algorithm/overload control/attitude control分类
计算机与自动化引用本文复制引用
谭富威,何永宁,孙晓晖,朱震,张庆昊,卢俊国..基于深度强化学习的飞行器过载和姿态智能控制研究[J].飞控与探测,2025,8(1):25-31,7.基金项目
中国航天科技集团有限公司第八研究院产学研合作基金(USCAST2022-34) （USCAST2022-34）
