基于神经网络PID算法的四旋翼无人机优化控制

doi:10.16180/j.cnki.issn1007-7820.2021.10.008

摘要/Abstract

摘要：

针对四旋翼无人机系统的时变性、非线性等问题,文中提出了一种基于神经网络PID的无人机姿态控制算法。该算法具有PID算法的强自适应能力和神经网络的强抗干扰能力。在建立四旋翼无人机刚体运动学模型和动力学模型基础上,搭建了四旋翼无人机MATLAB-Simulink姿态仿真控制模型,对比了神经网络PID控制算法与传统PID、串级PID控制算法对无人机3个姿态角的控制效果。结果表明,相比其他两种控制算法,神经网络PID控制算法调节过渡时间从4 s降到1 s,输出稳定后的系统静态误差只有0.5%,说明该控制算法控制精度高,具有更好的静态特性和动态特性。

关键词: 四旋翼, 无人机, 动力学建模, 仿真, 传统PID, 串级PID, 神经网络PID, 姿态控制

Abstract:

In this study, an attitude control algorithm based on neural network PID is proposed to solve the time-varying and nonlinear problems of the quadrotor UAVs system. The algorithm has the strong adaptive ability of PID algorithm and the strong anti-interference ability of neural network. Based on the establishment of the rigid body kinematics model and dynamics model of the quadrotor UAVs, the attitude simulation control model of the quadrotor UAV MATLAB-Simulink is built, and the control effect of the neural network PID control algorithm, traditional PID and cascade PID control algorithm on the three attitude angles of the UAVs is compared. The results show that compared with the other two control algorithms, the transition time of neural network PID control algorithm is reduced from 4 s to 1 s, and the static error of the system after output stabilization is only 0.5%, indicating that the neural network PID control algorithm has high control precision and better static and dynamic characteristics.

Key words: quadrotor, unmanned aerial vehicles, dynamics modeling, simulation, traditional PID, cascade PID, neural network PID, attitude control

中图分类号:

TP273

郭婕,金海,沈昕格. 基于神经网络PID算法的四旋翼无人机优化控制[J]. 电子科技, 2021, 34(10): 51-55.

GUO Jie,JIN Hai,SHEN Xinge. Optimization Control of Quadrotor UAVs Based on Neural Network PID Algorithm[J]. Electronic Science and Technology, 2021, 34(10): 51-55.

图/表 10

图1

表1

图2

图3

图4

表2

图5

表3

图6

图7

参考文献 18

[1]	杨则允, 李猛, 孙钦鹏. 四旋翼无人机控制系统仿真设计[J]. 计算机测量与控制, 2019, 27(4):68-71.
	Yang Zeyun, Li Meng, Sun Qinpeng. Simulation design of quad-rotor UAV control system[J]. Computer Measurement & Control, 2019, 27(4):68-71.
[2]	梅武军, 伍家成, 杨扬戬, 等. 一种小型无人机自主飞控系统设计与实现[J]. 电子科技, 2017, 30(7):106-109.
	Mei Wujun, Wu Jiacheng, Yang Yangjian, et al. A small unmanned aerial vehicle autonomous fight control system design[J]. Electronic Science and Technology, 2017, 30(7):106-109.
[3]	冯培晏. 四旋翼无人机建模与PID控制器设计[J]. 工业设计, 2018(6):135-137.
	Feng Peiyan. Quadrotor UAV modeling and PID controller design[J].Industrial Design, 2018(6):135-137.
[4]	梅武军. 基于ROS的多旋翼飞行器飞行控制系统的开发与设计[J]. 电子科技, 2017, 30(10):23-25.
	Mei Wujun. Development and design of flight control system for multi rotor vehicle based on ROS[J]. Electronic Science and Technology, 2017, 30(10):23-25.
[5]	郑健. 基于无模型自适应控制方法的四旋翼飞行器姿态调整[D]. 北京:北京交通大学, 2015.
	Zheng Jian. Quad-rotor aircraft attitude adjustment based on model-free adaptive control[D]. Beijing:Beijing Jiaotong University, 2015.
[6]	高青, 袁亮, 吴金强. 基于新型LQR的四旋翼无人机姿态控制[J]. 制造业自动化, 2014, 36(10):13-16.
	Gao Qing, Yuan Liang, Wu Jinqiang. Attitude control of a quadrotor UAV based on new LQR[J]. Manufacturing Automation, 2014, 36(10):13-16.
[7]	李珺, 王娜, 花玉, 等. 基于干扰观测器的四旋翼飞行器反步控制研究[J]. 计算机仿真, 2020, 37(4):28-33.
	Li Jun, Wang Na, Hua Yu, et al. Study on disturbance observer-based backstepping control for quadrotor[J]. Computer Simulation, 2020, 37(4):28-33.
[8]	张志赟, 谢习华. 四旋翼无人机模型参考自适应滑模控制[J]. 制造业自动化, 2020, 42(2):9-15.
	Zhang Zhiyun, Xie Xihua. Model reference adaptive sliding mode control for quadrotor UVA[J]. Manufacturing Automation, 2020, 42(2):9-15.
[9]	全权. 多旋翼飞行器设计与控制[M]. 北京: 电子工业出版社, 2018.
	Quan Quan. Multi-rotor aircraft design and control[M]. Beijing: Publishing House of Electronics Industry, 2018.
[10]	徐大远, 王英健, 陈冠军, 等. 四轴飞行器的动力学建模和位置控制研究[J]. 电子科技, 2015, 28(1):69-72.
	Xu Dayuan, Wang Yingjian, Chen Guanjun, et al. Research on dynamic modeling and position control of the quadcopter[J]. Electronic Science and Technology, 2015, 28(1):69-72.
[11]	Wu X W, Xiao B, Qu Y H. Modeling and sliding mode-based attitude tracking control of a quadrotor UAV with time-varying mass[J]. ISA Transactions, 2019(7):266-279.
[12]	方璇, 钟伯成. 基于人工鱼群PID控制算法的四旋翼飞行器控制[J]. 电子科技, 2015, 28(12):52-55.
	Fang Xuan, Zhong Bocheng. Four-rotor unmanned aerial vehicles control system based on PID controller of artificial fish swarm algorithm[J]. Electronic Science and Technology, 2015, 28(12):52-55.
[13]	徐会丽. 多旋翼无人机飞行控制算法研究[D]. 重庆:中国科学院大学, 2017.
	Xu Huili. Research on flight control algorithm of multi-rotor UAV[D]. Chongqing:University of Chinese Academy of Sciences, 2017.
[14]	温浩, 赵国庆. 基于MATLAB神经网络工具箱的线性神经网络实现[J]. 电子科技, 2005, 18(1):26-29.
	Wen Hao, Zhao Guoqing. Realization of the linear network based on the neural network tool kit in MATLAB[J]. Electronic Science and Technology, 2005, 18(1):26-29.
[15]	Paiva E, Llano M, Rodas J, et al. Design and implementati- on of a VTOL flight transition mechanism and development of a mathematical model for a tilt rotor UAV [C].Concepción:IEEE International Conference on Automation, 2018.
[16]	张硕, 张学典, 秦敏, 等. 基于Backstepping模糊自适应的四旋翼飞行器控制[J]. 电子科技, 2017, 30(2):54-57.
	Zhang Shuo, Zhang Xuedian, Qin Min, et al. Adaptive fuzzy backstepping control of quadrotors[J]. Electronic Science and Technology, 2017, 30(2):54-57.
[17]	熊中刚, 刘忠, 王寒迎, 等. RBF神经网络增量式PID自动转向控制系统设计[J]. 农机化研究, 2021, 43(4):27-32.
	Xiong Zhonggang, Liu Zhong, Wang Hanying, et al. Design of automatic steering control system based on RBF neural network incremental PID[J]. Journal of Agricultural Mechanization Research, 2021, 43(4):27-32.
[18]	黄旋, 陈光明, 赵潮. 基于STM32和μC/OS-Ⅲ的服务机器人设计与实现[J]. 机器人技术与应用, 2020(1):39-43.
	Huang Xuan, Chen Guangming, Zhao Chao. A service robot design based on the STM32 and μC/OS-Ⅲ[J]. Robot Technique and Application, 2020(1):39-43.

螺旋桨直径× 螺距/m	电池电压 /V	电流/A	油门量 /%	拉力/kg	功率/W
0.127 0× 0.114 3	12.6	1.59	29	0.10	20
		4.04	50	0.20	51
		7.10	66	0.30	89
		10.20	79	0.40	129
		13.76	91	0.50	173
		18.76	100	0.63	236

参数名	符号	数值	单位
无人机总质量	m	1.5	kg
无人机轴距	L	0.25	m
杭州重力加速度	g	9.793 6	m·s^-2
绕x轴的转动惯量	I_x	5.813×10^-3	kg·m²
绕y轴的转动惯量	I_y	5.813×10^-3	kg·m²
绕z轴的转动惯量	I_z	1.108×10^-2	kg·m²
机身阻力系数	K	2.336×10^-2	N·(m·s^-1)^-2

PITCH环PID参数	P参数	I参数	D参数
经典PID	2.00	0.40	0.70
串级PID内环	1.00	0.20	0.20
串级PID外环	1.50	0.50	0.30
神经网络PID	2.05	0.30	0.55