基于循环神经网络的四旋翼无人机路径跟踪控制  

Path Tracking Control of Quadrotor UAV Based on Recurrent Neural Network

作　　者：陆伟民 郑伟彦 黄迪 王宁涛 LU Weimin;ZHENG Weiyan;HUANG Di;WANG Ningtao(Hangzhou Science and Technology Development Branch,Zhejiang Dayou Industrial Co.,Ltd,Hangzhou 310051)

机构地区：[1]浙江大有实业有限公司杭州科技发展分公司,杭州310051

出　　处：《导航与控制》2025年第1期90-98,共9页Navigation and Control

基　　金：浙江大有集团有限公司科技项目(编号:DY2023-09)。

摘　　要：四旋翼无人机飞行时,通常易遭受不确定环境因素影响,导致控制系统易出现模型不确定和内部未知扰动问题。针对四旋翼无人机遭受模型不确定和内部未知扰动情形下的路径跟踪控制进行研究,提出了基于循环神经网络的自适应滑模控制策略,利用自适应滑模控制策略确保路径跟踪误差渐近收敛。考虑系统集总干扰项,通过循环神经网络估计补偿系统集总干扰项,并引入自适应控制技术对循环神经网络组合逼近误差进行估计补偿,避免组合逼近误差对系统稳定性分析造成复杂计算而影响路径跟踪性能。仿真结果表明,所提控制策略具备有效性和鲁棒性,同时具有良好的路径跟踪控制性能。During the flight of quadrotor unmanned aerial vehicle UAV,it is often susceptible to uncertain environmental factors,leading to model uncertainties and unknown internal disturbances in the control system.A study is conducted on the path tracking control of quadrotor UAV under model uncertainty and internal unknown disturbances.An adaptive sliding mode control strategy based on recurrent neural networks is proposed for ensuring that the path tracking errors gradually converge using the adaptive sliding mode control strategy.Considering the lumped disturbance terms of the system,a recurrent neural network is used to estimate and compensate for the lumped disturbance terms,and adaptive control technology is introduced to estimate and compensate for the combined approximation errors of recurrent neural network,avoiding the complex calculation caused by the combined approximation errors on the system stability analysis and affecting the path tracking performance.The simulation results show that the control strategy proposed in this paper is effective and robust,and has good path tracking control performance.

关 键 词：四旋翼无人机 循环神经网络 自适应滑模控制 路径跟踪控制 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]