1 College of Electromechanical Engineering, China Jiliang University, Hangzhou 310008, China
2 College of Mechanical Engineering, Zhejiang University, Hangzhou 310008, China
Show less
文章历史+
Received
Published
Revised
2025-07-31
2026-08-18
2026-01-09
Online Date
2026-09-04
PDF (10685K)
摘要
为提升气动肌肉(pneumatic artificial muscle,PAM)拮抗关节系统的轨迹跟踪精度,提出一种基于BP神经网络(back propagation neural network,BPNN)的预测控制方法。首先在气动肌肉拮抗系统动力学分析的基础上,通过BP神经网络建模,构建融合误差角度与误差角速度的PAM拮抗系统神经网络模型,并对其进行离散化处理以建立预测模型。其次,基于该预测模型设计BP神经网络模型预测控制器(BPNN-based model predictive controller,BPNNMPC),通过滚动优化实现关节角度的高精度预测控制。最后为验证控制方法的优越性,开展仿真与实物实验的对比验证。结果表明:所提出的BPNNMPC能有效提升轨迹跟踪性能。验证了所提气动肌肉控制方法的有效性。
Abstract
To improve the trajectory tracking accuracy of the pneumatic artificial muscle (PAM) antagonistic joint system, a predictive control method based on a backpropagation neural network (BPNN) was proposed. First, based on the dynamic analysis of the PAM antagonistic system, a neural network model integrating the error angle and error angular velocity was constructed using BPNN and discretized to establish a predictive model. Second, a BPNN-based model predictive controller (BPNNMPC) was designed based on this predictive model, by which high-precision joint angle control was achieved through rolling optimization. Finally, comparative simulations and physical experiments were conducted to validate the superiority of the proposed method. It is demonstrated by the results demonstrate that the trajectory tracking performance is effectively enhanced by the BPNNMPC.
ZONG Qing-tao, WANG Bin-rui, CHENG Tai, ZHANG Bin, LIU Dan, LIU Tao, FANG Jun-wei, ZHOU Kun.
Predictive Control of Pneumatic Artificial Muscle Antagonist Joint Angle Tracking Model Based on BP Neural Network Modeling[J].
Science Technology and Engineering, 2026, 26(23): 10031-10040 DOI:10.12404/j.issn.1671-1815.2505642
针对上述挑战,现提出一种基于BP神经网络建模的气动人工肌肉拮抗关节角度跟踪模型预测控制方法。首先,在建立PAM拮抗关节非线性动力学模型的基础上,利用BP神经网络算法对模型关键参数进行辨识,并推导得到系统的非线性预测模型;其次,基于该预测模型设计BP神经网络模型预测控制器(BPNN-based model predictive controller,BPNNMPC),通过滚动优化求解最优控制律;最后,通过仿真分析与实物平台实验,将所提算法与未采用神经网络优化的模型预测控制进行对比。提出的BP NNMPC能够有效建立高精度的预测模型,用于提升PAM拮抗关节的轨迹跟踪性能。
ChenGaoming, LuoYan, HuangBiyi, et al. Robot calibration-free visual servo technology based on BP neural network-fuzzy control[J]. Science Technology and Engineering, 2023, 23(26): 11282-11291.
[3]
DellaS C, DuriezC, RusD. Model-based control of soft robots: a survey of the state of the art and open challenges[J]. IEEE Control Systems Magazine, 2023, 43(3): 30-65.
[4]
LiL B, ZhuL J, HuangJ. Stable locomotion of biped robot with gaits of sinusoidal harmonics[J]. IEEE Transactions on Control Systems Technology, 2024, 32(3): 805-817.
[5]
ZhuK, LiB, ZheW M, et al. Collision avoidance among dense heterogeneous agents using deep reinforcement learning[J]. IEEE Robotics and Automation Letters, 2023, 8(1): 57-64.
[6]
XieD S, LiuJ B, ZuoS Y. Pneumatic artificial muscle with large stroke based on a contraction ratio amplification mechanism and self-contained sensing[J]. IEEE Robotics and Automation Letters, 2021, 6(4): 8599-8606.
[7]
DiaoS Z, LiuG D, LiuZ Q, et al. Prescribed-time adaptive fuzzy control for pneumatic artificial muscle actuated parallel robots with input constraints[J]. IEEE Transactions on Fuzzy Systems, 2024, 32(4): 2039-2051.
[8]
ZhangY, ChengQ, ChenW L, et al. Dynamic modeling and inverse compensation for coupled hysteresis in pneumatic artificial muscle-actuated soft manipulator with variable stiffness[J]. ISA Transactions, 2024, 145: 468-478.
[9]
WangB, WangY, HuangJ, et al. Computed torque control and force analysis for mechanical leg with variable rotation axis powered by servo pneumatic muscle[J]. ISA Transactions, 2023, 140: 385-401.
[10]
TanakaS, NabaeH, SuzumoriK. Back-stretchable McKibben muscles: expanding the range of antagonistic muscle driven joints[J]. IEEE Robotics and Automation Letters, 2023, 8(9): 5331-5337.
[11]
WangX Y, YuN B, HanJ D, et al. Modeling and robust control for tendon-sheath artificial muscle system twist with time-varying parameters and input constraints[J]. IEEE Transactions on Industrial Electronics, 2023, 70(1): 878-887.
[12]
XiY G, LiD W, LinS. Model predictive control-status and challenges[J]. Acta Automatica Sinica, 2013, 39(3): 222-236.
[13]
PanG, FaulwasserT. NMPC in active subspaces: dimensionality reduction with recursive feasibility guarantees[J]. Automatica, 2023, 147: 110708.
[14]
HuangL B, ZhenJ Z, LygerosJ, et al. Robust data-enabled predictive control: tractable formulations and performance guarantees[J]. IEEE Transactions on Automatic Control, 2023, 68(5): 3163-3170.
[15]
ChenS W, WangT Y, AtanasovN, et al. Large scale model predictive control with neural networks and primal active sets[J]. Automatica, 2019, 135: 109947.
[16]
TerziE, FagianoL, FarinaM, et al. Robust multi-rate predictive control using multi-step prediction models learned from data[J]. Automatica, 2021, 136: 109852.
ChengTai, WangBinrui, ZongQingtao, et al. Model predictive control of pneumatic muscle driven three-fingered dexterous hand based on radial basis function neural network[J/OL]. (2025-07-03)[2025-10-31]. https://doi.org/10.13196/j.cims.2025.0308.
[19]
WangJ J, XuB G, LaiJ W, et al. An improved Koopman-MPC framework for data-driven modeling and control of soft actuators[J]. IEEE Robotics and Automation Letters, 2022, 8(2): 616-623.
[20]
CaoY, HuangJ, XiongC H. Single-layer learning-based predictive control with echo state network for pneumatic-muscle-actuators driven exoskeleton[J]. IEEE Transactions on Cognitive and Developmental Systems, 2021, 13(1): 80-90.
DuHongwang, XuXiaoya, ShaoRenbo, et al. Transient dynamic modeling, analysis and validation of pneumatic bending artificial muscle[J]. Journal of Mechanical Engineering, 2023, 59(21): 199-208.
[23]
ChenY H, SunN, LiangD K, et al. A neuroadaptive control method for pneumatic artificial muscle systems with hardware experiments[J]. Mechanical Systems and Signal Processing, 2021, 146: 106976.
[24]
DuongM D, PhamQ T, VuT C, et al. Adaptive fuzzy sliding mode control of an actuator powered by two opposing pneumatic artificial muscles[J]. Scientific Reports, 2023, 13: 8242.
[25]
ChenC T, LienW Y, WuY C. Implementation of an upper-limb exoskeleton robot driven by pneumatic muscle actuators for rehabilitation[J]. Actuators, 2020, 9(4): 106.
ShengJiahao, LiuLi, LiuZhaohui, et al. Asphalt mixture performance prediction method based on genetic algorithm-optimized BP neural network[J]. Science Technology and Engineering, 2025, 25(3): 1214-1224.
[28]
余芮. 基于自适应反步控制的气动肌肉关节轨迹追踪研究[D]. 杭州: 中国计量大学, 2020.
[29]
YuRui. Trajectory tracking research of pneumatic muscle joint based on adaptive backstepping control[D]. Hangzhou: China Jiliang University, 2020.
ZhangChuanqiang, ZhouKun, CuiXiaohong, et al. Predictive angle control for pneumatic muscle antagonistic joint based on T-S fuzzy modeling[J]. Journal of China University of Metrology, 2024, 35(2): 222-232.