文献类型：期刊文献

中文题名：基于自适应RBF神经网络具有模型不确定性的四旋翼无人机指定时间预设性能控制方法

英文题名：Adaptive RBF Neural Networks for Appointed-time Performance Control of Quadcopter UAVs with Model Uncertainty

作者：张园[1,2];郑鸿基[1,3];刘海涛[3];韦丽娇[1,4];沈德战[1,5];赵振华[1,5]

机构：[1]中国热带农业科学院农业机械研究所,湛江524091;[2]广东省农业类颗粒体精量排控工程技术研究中心,湛江524000;[3]广东海洋大学机械工程学院,湛江524091;[4]湛江市类颗粒体动力学及精准精量排控重点实验室,湛江524091;[5]农业农村部热带作物农业装备重点实验室,湛江524091

年份：2024

卷号：55

期号：4

起止页码：64

中文期刊名：农业机械学报

外文期刊名：Transactions of the Chinese Society for Agricultural Machinery

收录：北大核心2023、CSTPCD、、EI(收录号：20242116113050)、Scopus、CSCD2023_2024、北大核心、CSCD

基金：海南省重点研发计划项目(ZDYF2024XDNY152);广东省企业科技特派员专项(GDKTP2021008500);湛江市科技计划项目(2022A105&2020A05004&2021A05194);广东省教育厅重点项目(2021ZDZX1041);深圳市科技计划项目(JCYJ20220530162014033)。

语种：中文

中文关键词：四旋翼无人机;RBF神经网络;轨迹跟踪控制;预设性能约束;模型不确定性

外文关键词：quadrotor UAV;RBF neural network;trajectory tracking control;prescribed performance control;model uncertainty

中文摘要：四旋翼无人机具有强耦合和欠驱动的特点,在飞行过程中很容易受到外界干扰,进而影响整个无人机系统的稳定性和精度。为此,提出了一种基于RBF神经网络的指定时间预设性能约束控制策略。首先,针对四旋翼无人机的不确定数学模型难以精确建立,并且在执行任务过程中存在外部未知扰动问题,提出了一种基于指定时间预设性能控制方法,将四旋翼无人机的轨迹跟踪问题转换为对位置子系统和姿态子系统的期望指令跟踪问题;其次,在设计控制器过程中,为了解决“微分爆炸”问题产生的滤波器误差,引入一种新型滤波误差补偿方法,通过RBF神经网络逼近外部未知扰动,并将预测结果补偿给控制器以提高轨迹跟踪的鲁棒性。最后,应用仿真模拟方法验证无人机控制系统稳定性和性能优势,通过飞行试验验证,微风聚拢环境下实际飞行轨迹与仿真模拟结果趋于一致,自主轨迹跟踪起降位置偏差小于1 cm,证明了所提出算法的有效性。

外文摘要：Quadrotor UAVs are characterized by strong coupling and underdrive,and are easily affected by external interference during flight,which in turn affects the stability and accuracy of the whole UAV system.Aiming at this problem,a specified-time preset performance constraint control policy based on RBF neural network was proposed.Firstly,in view of the difficulty of establishing an accurate mathematical model for the uncertain mathematical model of the quadrotor UAV and the existence of external unknown disturbances during the execution of the mission,a control method based on the specified time preset performance constraints was proposed,and the trajectory tracking problem of the quadrotor UAV was transformed into the desired command tracking problem for the position subsystem and the attitude subsystem;in view of the design of the controller,in order to solve the problem of the “position subsystem”,the RBF neural network was used to design the controller.Secondly,a compensation system was introduced to solve the filter error caused by the “differential explosion” problem during the controller design process.Finally,the unknown external perturbations were compensated by RBF neural network approximation and the predicted results were compensated to the controller to improve the robustness.Finally,the simulation method is used to verify the stability and performance advantages of UAV control system,flight tests were conducted to verify that the actual flight trajectory in a breeze gathering environment tended to be consistent with the simulation results.The deviation of the autonomous trajectory tracking takeoff and landing position was less than 1 cm,demonstrating the effectiveness of the proposed algorithm.