基于神经网络ODE和非线性MPC的DEA建模与控制

黄鹏; 王亚午; 吴俊东; 苏春翌; 福岛E.文彦

doi:10.16383/j.aas.c240223

基于神经网络ODE和非线性MPC的DEA建模与控制

doi: 10.16383/j.aas.c240223

黄鹏^{1, 2, 3, 4,},
王亚午^{1, 2, 3,},
吴俊东^{1, 2, 3,},
苏春翌^5,,
福岛E.文彦^4,

1.
中国地质大学(武汉)自动化学院武汉 430074 中国
2.
复杂系统先进控制与智能自动化湖北省重点实验室武汉 430074 中国
3.
地球探测智能化技术教育部工程研究中心武汉 430074 中国
4.
东京工科大学工学部东京 192-0982 日本
5.
康考迪亚大学Gina Cody工程与计算机科学学院蒙特利尔 QC H3G 1M8 加拿大

基金项目: 国家自然科学基金面上项目(62273316), 国家建设高水平大学公派研究生项目(202206410064), 高等学校学科创新引智计划项目(111计划)(B17040)资助

详细信息

作者简介:
黄鹏：中国地质大学(武汉)自动化学院博士研究生. 主要研究方向为软体机器人和机器人控制. E-mail: huangpeng@cug.edu.cn

王亚午：中国地质大学(武汉)自动化学院教授. 主要研究方向为机器人控制和非线性系统控制. 本文通信作者. E-mail: wangyawu@cug.edu.cn

吴俊东：中国地质大学(武汉)自动化学院教授. 主要研究方向为软体机器人和非线性系统控制. E-mail: jdwu@cug.edu.cn

苏春翌：加拿大康考迪亚大学Gina Cody工程与计算机科学学院教授. 主要研究方向为机器人控制, 非线性系统控制和软体机器人. E-mail: chun-yi.su@concordia.ca

福岛E.文彦：日本东京工科大学工学部教授. 主要研究方向为机器人和机电一体化. E-mail: fukushimafmhk@stf.teu.ac.jp

计量
- 文章访问数: 129
- HTML全文浏览量: 67
- 被引次数: 0
出版历程
- 收稿日期: 2024-04-24
- 录用日期: 2024-08-14
- 网络出版日期: 2024-09-06

Modeling and Control of Dielectric Elastomer Actuator Based on Neural Ordinary Differential Equation and Nonlinear Model Predictive Control

HUANG Peng^{1, 2, 3, 4
,},
WANG Ya-Wu^{1, 2, 3
,},
WU Jun-Dong^{1, 2, 3
,},
SU Chun-Yi^5
,,
FUKUSHIMA Edwardo-Fumihiko^4
,

1.
School of Automation, China University of Geosciences, Wuhan 430074, China
2.
Hubei Key Laboratory of Advanced Control and Intelligent Automation for Complex Systems, Wuhan 430074, China
3.
Engineering Research Center of Intelligent Technology for Geo-Exploration, Ministry of Education, Wuhan 430074, China
4.
School of Engineering, Tokyo University of Technology, Tokyo 192-0982, Japan
5.
Gina Cody School of Engineering and Computer Science, Concordia University, Montreal QC H3G 1M8, Canada

Funds: Supported by General Program of National Natural Science Foundation of China (62273316), Program of China Scholarship Council (202206410064), and Higher Education Discipline Innovation Project of China (111 Project) (B17040)

More Information

Author Bio:
HUANG Peng　Ph.D. candidate at the School of Automation, China University of Geosciences. His research interest covers soft robotics and robot control

WANG Ya-Wu　Professor at the School of Automation, China University of Geosciences. His research interest covers robot control and nonlinear system control. Corresponding author of this paper

WU Jun-Dong　Professor at the School of Automation, China University of Geosciences. His research interest covers soft robotics and nonlinear system control

SU Chun-Yi　Professor at the Gina Cody School of Engineering and Computer Science, Concordia University. His research interest covers robot control, nonlinear system control and soft robotics

FUKUSHIMA Edwardo-Fumihiko　Professor at the School of Engineering, Tokyo University of Technology. His research interest covers robotics and mechatronics

摘要

摘要: 针对介电弹性体驱动器(Dielectric elastomer actuator, DEA)建模与控制的挑战性问题, 提出基于神经网络常微分方程(Ordinary differential equation, ODE)和非线性模型预测控制(Model predictive control, MPC)的DEA动力学建模与跟踪控制方法. 首先, 基于神经网络ODE建立DEA的动力学模型以描述其复杂的动态行为. 然后, 基于所建立的DEA动力学模型, 设计非线性模型预测控制器实现其跟踪控制目标. 最后, 在所搭建的实验平台上进行一系列跟踪控制实验. 在所有实验结果中, DEA的运动均能很好地跟踪目标轨迹, 且相对均方根误差均不超过3.30%, 说明了所提动力学建模与跟踪控制方法的有效性.
- 介电弹性体驱动器 /
- 神经网络常微分方程 /
- 动力学建模 /
- 非线性模型预测控制
Abstract: This paper proposed dynamic modeling and tracking control methods for a dielectric elastomer actuator (DEA) based on the neural ordinary differential equation (ODE) and nonlinear model predictive control (MPC). First, a dynamic model of the DEA was established based on the neural ODE to describe its complicated dynamics behavior. Then, based on the established dynamic model of the DEA, a nonlinear model predictive controller was designed to realize its tracking control objective. Finally, a series of tracking control experiments were conducted on the built experimental platform. In all experimental results, the motion of the DEA can track the target trajectory well, and all relative root-mean-square-errors are no more than 3.30%, which illustrates the effectiveness of proposed dynamic modeling and tracking control methods.
- Dielectric elastomer actuator /
- neural ordinary differential equation /
- dynamic modeling /
- nonlinear model predictive control

HTML全文

图 1 锥形介电弹性体驱动器

Fig. 1 Conical dielectric elastomer actuator

下载: 全尺寸图片幻灯片

图 2 实验平台

Fig. 2 Experimental platform

下载: 全尺寸图片幻灯片

图 3 动力学模型结构框图

Fig. 3 Structure diagram of dynamic model of DEA

下载: 全尺寸图片幻灯片

图 4 神经网络结构图

Fig. 4 Structure diagram of neural network

下载: 全尺寸图片幻灯片

图 5 损失函数随迭代次数的变化曲线

Fig. 5 Curve of loss function with iterations

下载: 全尺寸图片幻灯片

图 6 DEA动力学模型输出值与实验测量值的对比

Fig. 6 Comparison between output of dynamic model of DEA and experimental measurement

下载: 全尺寸图片幻灯片

图 7 在不同驱动电压幅值情况下的模型验证结果

Fig. 7 Model validation results in different actuation voltage amplitudes

下载: 全尺寸图片幻灯片

图 8 在不同驱动电压频率情况下的模型验证结果

Fig. 8 Model validation results in different actuation voltage frequencies

下载: 全尺寸图片幻灯片

图 9 控制系统整体结构框图

Fig. 9 Structure diagram of whole control system

下载: 全尺寸图片幻灯片

图 10 对正弦波的跟踪结果

Fig. 10 Tracking result for sinusoidal wave

下载: 全尺寸图片幻灯片

图 11 对三角波的跟踪结果

Fig. 11 Tracking result for triangular wave

下载: 全尺寸图片幻灯片

图 12 对阶梯波的跟踪结果

Fig. 12 Tracking result for step wave

下载: 全尺寸图片幻灯片

图 13 对复合波的跟踪结果

Fig. 13 Tracking result for composite wave

下载: 全尺寸图片幻灯片

表 1 第1个实验方案中所有实验的相对均方根误差

Table 1 $E_R$ for all experiments of first experimental scheme

$m$	$a_m$ (kV)	$E_R$
1	5.0	3.12%
2	6.0	1.53%
3	7.0	1.55%
4	8.0	2.16%

下载: 导出CSV

表 2 第2个实验方案中所有实验的相对均方根误差

Table 2 $E_R$ for all experiments of second experimental scheme

$m$	${\psi}_m$ (Hz)	$E_R$
1	0.2	1.82%
2	0.6	2.10%
3	1.0	2.50%
4	1.4	1.89%

下载: 导出CSV

表 3 ${{N}_{p}}$和${{N}_{c}}$取值对控制精度、单步运行时间和实时性的影响

Table 3 Influences of values of ${{N}_{p}}$ and ${{N}_{c}}$ on control accuracy, single-step running time and real-time performance

${{N}_{p}}$	${{N}_{c}}$	$E_R$	$T_c$/s	实时性
2	1	3.26%	$5.08\times {10^{ - 3}}$	满足
4	2	2.83%	$8.39\times {10^{ - 3}}$	满足
5	2	1.88%	$8.91\times {10^{ - 3}}$	满足
6	4	—	$1.85\times {10^{ - 2}}$	不满足

下载: 导出CSV

参考文献(28)

[1]	Rus D, Tolley M T. Design, fabrication and control of soft robots. Nature, 2015, 521(7553): 467−475 doi: 10.1038/nature14543
[2]	文力, 王贺升. 软体机器人研究展望: 结构、驱动与控制. 机器人, 2018, 40(5): 577 Wen Li, Wang He-Sheng. Prospects for soft robotics research: structure, actuation and control. Robot, 2018, 40(5): 577
[3]	徐璠, 王贺升. 软体机械臂水下自适应鲁棒视觉伺服. 自动化学报, 2023, 49(4): 744−753 Xu Fan, Wang He-Sheng. Adaptive robust visual servoing control of a soft manipulator in underwater environment. Acta Automatica Sinica, 2023, 49(4): 744−753
[4]	Naclerio N D, Karsai A, Cooper M M. Controlling subterranean forces enables a fast, steerable, burrowing soft robot. Science Robotics, 2021, 6(55): Article No. eabe2922 doi: 10.1126/scirobotics.abe2922
[5]	王成军, 邓海龙. 软体机器人在康复训练领域的应用. 机械传动, 2024, 48(5): 169−176 Wang Cheng-Jun, Deng Hai-Long. Application of soft robots in rehabilitation training. Journal of Mechanical Transmission, 2024, 48(5): 169−176
[6]	李兴旺, 滕燕, 徐迎. 一种基于双稳态驱动器的气动软体四足机器人. 机器人, 2024, 46(3): 294−304 Li Xing-Wang, Teng Yan, Xu Ying. A pneumatic soft quadruped robot based on a bistable actuator. Robot, 2024, 46(3): 294−304
[7]	Wu J D, Ye W J, Wang Y W, Su C Y. Modeling based on a two-step parameter identification strategy for liquid crystal elastomer actuator considering dynamic phase transition process. IEEE Transactions on Cybernetics, 2023, 53(7): 4423−4434 doi: 10.1109/TCYB.2022.3179433
[8]	Pelrine R, Kornbluh R, Pei Q B, Joseph J. High-speed electrically actuated elastomers with strain greater than 100%. Science, 2000, 287(5454): 836−839 doi: 10.1126/science.287.5454.836
[9]	李智, 陈国强, 徐泓智, 陈新开, 陕晋军, 张秀宇. 介电弹性体驱动系统建模及控制方法综述. 控制与决策, 2023, 38(8): 2283−2300 Li Zhi, Chen Guo-Qiang, Xu Hong-Zhi, Chen Xin-Kai, Shan Jin-Jun, Zhang Xiu-Yu. A review of modeling and control methods for dielectric elastomer actuator systems. Control and Decision, 2023, 38(8): 2283−2300
[10]	Chen Y F, Zhao H C, Mao J, Chirarattananon P, Helbling E F, Hyun N P, et al. Controlled flight of a microrobot powered by soft artificial muscles. Nature, 2019, 575(7782): 324−329 doi: 10.1038/s41586-019-1737-7
[11]	Gu G Y, Zou J, Zhao R K, Zhao X H, Zhu X Y. Soft wall-climbing robots. Science Robotics, 2018, 3(25): Article No. eaat2874 doi: 10.1126/scirobotics.aat2874
[12]	Li G R, Chen X P, Zhou F H, Liang Y M, Xiao Y H, Cao X N, et al. Self-powered soft robot in the Mariana Trench. Nature, 2021, 591(7848): 66−71 doi: 10.1038/s41586-020-03153-z
[13]	Zhang Y, Wu J D, Meng Q X, Wang Y W, Su C Y. Robust control of dielectric elastomer smart actuator for tracking high-frequency trajectory. IEEE Transactions on Industrial Informatics, 2024, 20(1): 224−234 doi: 10.1109/TII.2023.3258445
[14]	Huang P, Wu J D, Zhang P, Wang Y W, Su C Y. Dynamic modeling and tracking control for dielectric elastomer actuator with a model predictive controller. IEEE Transactions on Industrial Electronics, 2022, 69(2): 1819−1828 doi: 10.1109/TIE.2021.3063976
[15]	Suo Z G. Theory of dielectric elastomers. Acta Mechanica Solida Sinica, 2010, 23(6): 549−578 doi: 10.1016/S0894-9166(11)60004-9
[16]	Xiao H, Wu J D, Ye W J, Wang Y W. Dynamic modeling of dielectric elastomer actuators based on thermodynamic theory. Mechanics of Advanced Materials and Structures, 2022, 29(11): 1543−1552 doi: 10.1080/15376494.2020.1829757
[17]	Huang P, Ye W J, Wang Y W. Dynamic modeling of dielectric elastomer actuator with conical shape. Plos One, 2020, 15(8): Article No. e0235229 doi: 10.1371/journal.pone.0235229
[18]	Zou J, Gu G Y. High-precision tracking control of a soft dielectric elastomer actuator with inverse viscoelastic hysteresis compensation. IEEE/ASME Transactions on Mechatronics, 2019, 24(1): 36−44 doi: 10.1109/TMECH.2018.2873620
[19]	Wang H C, Fu T F, Du Y Q, Gao W H, Huang K X, Liu Z M, et al. Scientific discovery in the age of artificial intelligence. Nature, 2023, 620(7972): 47−60 doi: 10.1038/s41586-023-06221-2
[20]	Linot A J, Burby J W, Tang Q, Balaprakash P, Graham M D, Maulik R. Stabilized neural ordinary differential equations for long-time forecasting of dynamical systems. Journal of Computational Physics, 2023, 474: Article No. 111838 doi: 10.1016/j.jcp.2022.111838
[21]	Hornik K, Stinchcombe M, White H. Universal approximation of an unknown mapping and its derivatives using multilayer feedforward networks. Neural Networks, 1990, 3(5): 551−560 doi: 10.1016/0893-6080(90)90005-6
[22]	Gupta U, Wang Y Z, Ren H L, Zhu J. Dynamic modeling and feedforward control of jaw movements driven by viscoelastic artificial muscles. IEEE/ASME Transactions on Mechatronics, 2019, 24(1): 25−35 doi: 10.1109/TMECH.2018.2875521
[23]	Wang Y W, Huang P, Wu J D, Su C Y. Modelling and compound control of intelligently dielectric elastomer actuator. Control Engineering Practice, 2022, 126: Article No. 105261 doi: 10.1016/j.conengprac.2022.105261
[24]	Lu F, Lin K K, Chorin A J. Comparison of continuous and discrete-time data-based modeling for hypoelliptic systems. Communications in Applied Mathematics and Computational Science, 2016, 11(2): 187−216 doi: 10.2140/camcos.2016.11.187
[25]	陈虹. 模型预测控制. 北京: 科学出版社, 2013. 245−251 Chen Hong. Model Predictive Control. Beijing: Science Press, 2013. 245−251
[26]	席裕庚, 李德伟, 林姝. 模型预测控制—现状与挑战. 自动化学报, 2013, 39(3): 222−236 doi: 10.1016/S1874-1029(13)60024-5 Xi Yu-Geng, Li De-Wei, Lin Shu. Model predictive control — Status and challenges. Acta Automatica Sinica, 2013, 39(3): 222−236 doi: 10.1016/S1874-1029(13)60024-5
[27]	章越. 介电弹性体驱动器动力学建模与轨迹跟踪控制 [硕士学位论文]. 中国地质大学, 2023 Zhang Yue. Dynamic modeling and trajectory tracking control of dielectric elastomer actuator [Master thesis]. China University of Geosciences, China, 2023
[28]	Zhang X Y, Xu H Z, Chen X K, Li Z, Su C Y. Modeling and adaptive output feedback control of butterfly-like hysteretic nonlinear systems with creep and their applications. IEEE Transactions on Industrial Electronics, 2023, 70(5): 5182−5191 doi: 10.1109/TIE.2022.3187583