RBF神经网络补偿的并联机器人控制研究
|
摘要
为了实现对三自由度Delta并联机器人更精确的轨迹跟踪控制,对并联机构的动力学建模不确定性进行研究,提出了计算力矩控制基础上的RBF神经网络在线补偿控制策略。利用Lyapunov理论推导了神经网络在线权值自适应律,保证了系统稳定性。运用RBF神经网络在线自学习系统的不确定性,提高了控制效率同时增加算法的自适应性。在Simmechanics中建立系统物理模型并在Simulink中设计控制器,之后进行Simulimk/Simmechanics联合仿真,结果表明算法优于计算力矩控制,可以有效减小跟踪误差的收敛半径,实现对目标轨迹的准确跟踪。
In order to realize a more accurate trajectory tracking control of the 3-DOF delta parallel robot,the uncertainty of parallel mechanism's dynamic model was researched,a control strategy based on on-line error compensated by RBF neural network was proposed. The law of network weights was develop ed based on Lyapunov theory and the system stability was guaranteed. The system instability was learned on-line by RBF neural network,the control efficiency and the adaptability of algorithm were improved. The physical model was established in Simmechanics and the controller was designed in Simulink.Then the co-simulation was realized based on Simulink/Simmechanics. The results show that the algorithm is better than computed torque control. It can reduce errors effectively and realize the precise track of target trajectory.
In order to realize a more accurate trajectory tracking control of the 3-DOF delta parallel robot,the uncertainty of parallel mechanism's dynamic model was researched,a control strategy based on on-line error compensated by RBF neural network was proposed. The law of network weights was develop ed based on Lyapunov theory and the system stability was guaranteed. The system instability was learned on-line by RBF neural network,the control efficiency and the adaptability of algorithm were improved. The physical model was established in Simmechanics and the controller was designed in Simulink.Then the co-simulation was realized based on Simulink/Simmechanics. The results show that the algorithm is better than computed torque control. It can reduce errors effectively and realize the precise track of target trajectory.
引文
[1]宫赤坤,蓝黎恩.Delta并联机器人的动力学研究[J].机械科学与技术,2013(12):1780-1784.(Gong Chi-kun,Lan Li-en.Study on the dynamics analysis of delta parallel robot[J].Mechanical Science and Technology,2013(12):1780-1784.)
[2]申铁龙.机器人鲁棒控制基础[M].北京:清华大学出版社,2000.(Shen Tie-long.Robust Control of Robot[M].Beijing:Tsinghua University Press,2000.)
[3]杨晓钧,龙亿.计算力矩法的CMAC同步轨迹跟踪控制与仿真[J].哈尔滨工业大学学报,2013(7):85-89.(Yang Xiao-jun,Long Yi.Synchronous trajectory tracking control and simulation of CMAC neural network based on computed torque control[J].Journal of Harbin Institute of Technology,2013(7):85-89.)
[4]贺红林,何文丛,刘文光.神经网络与计算力矩复合的机器人运动轨迹跟踪控制[J].农业机械学报,2013(5):270-275.(He Hong-lin,He Wen-cong,Liu Wen-guang.Tracking control of robot using hybrid controller based on neural network and computed torque[J].Transactions of the Chinese Society for Agricultural Machinery,2013(5):270-275.)
[5]Linda O,Manic M.Evaluating uncertainty resiliency of type2 fuzzy logic controllers for parallel Delta robot[C].IEEE 4th International Conference on Human System Interactions.Piscataway,USA:IEEE,201(5):91-97.
[6]江道根,高国琴,胡红玉.六自由度并联机器人线性化反馈RBF神经滑模控制研究[J].机械设计与制造,2010(2):159-161.(Jiang Dao-gen,Gao Guo-qin,Hu Hong-yu.The study on parallel robot of RBF netrual sliding controller[J].Machinery Design&Manufacture,2010(2):159-161.)
[7]Park J,Sandberg L W.Universal approximation using radial-basis-function networks[J].Neutral Comput,1991,3(2):246-257.
[8]刘金琨.机器人控制系统的设计与MATLAB仿真[M].北京:清华大学出版社,2008.(Liu Jin-kun.Design and MATLAB Simulation of Roboticsystems[M].Beijing:Tsinghua University Press,2008.)
[9]Katsuhiko Ogata.Modern Control Engineering[M].USA:Prentice Hall,2007.
[10]王英波,黄其涛,郑书涛.Simulink和Sim Mechanics环境下并联机器人动力学建模与分析[J].哈尔滨工程大学学报,2012(1):100-105.(Wang Ying-bo,Huang Qi-tao,Zheng Shu-tao.Dynamic modeling and analysis of a parallel manipulator using Simulink and Sim Mechanics[J].Journal of Harbin Enginering University,2012(1):100-105.)
[2]申铁龙.机器人鲁棒控制基础[M].北京:清华大学出版社,2000.(Shen Tie-long.Robust Control of Robot[M].Beijing:Tsinghua University Press,2000.)
[3]杨晓钧,龙亿.计算力矩法的CMAC同步轨迹跟踪控制与仿真[J].哈尔滨工业大学学报,2013(7):85-89.(Yang Xiao-jun,Long Yi.Synchronous trajectory tracking control and simulation of CMAC neural network based on computed torque control[J].Journal of Harbin Institute of Technology,2013(7):85-89.)
[4]贺红林,何文丛,刘文光.神经网络与计算力矩复合的机器人运动轨迹跟踪控制[J].农业机械学报,2013(5):270-275.(He Hong-lin,He Wen-cong,Liu Wen-guang.Tracking control of robot using hybrid controller based on neural network and computed torque[J].Transactions of the Chinese Society for Agricultural Machinery,2013(5):270-275.)
[5]Linda O,Manic M.Evaluating uncertainty resiliency of type2 fuzzy logic controllers for parallel Delta robot[C].IEEE 4th International Conference on Human System Interactions.Piscataway,USA:IEEE,201(5):91-97.
[6]江道根,高国琴,胡红玉.六自由度并联机器人线性化反馈RBF神经滑模控制研究[J].机械设计与制造,2010(2):159-161.(Jiang Dao-gen,Gao Guo-qin,Hu Hong-yu.The study on parallel robot of RBF netrual sliding controller[J].Machinery Design&Manufacture,2010(2):159-161.)
[7]Park J,Sandberg L W.Universal approximation using radial-basis-function networks[J].Neutral Comput,1991,3(2):246-257.
[8]刘金琨.机器人控制系统的设计与MATLAB仿真[M].北京:清华大学出版社,2008.(Liu Jin-kun.Design and MATLAB Simulation of Roboticsystems[M].Beijing:Tsinghua University Press,2008.)
[9]Katsuhiko Ogata.Modern Control Engineering[M].USA:Prentice Hall,2007.
[10]王英波,黄其涛,郑书涛.Simulink和Sim Mechanics环境下并联机器人动力学建模与分析[J].哈尔滨工程大学学报,2012(1):100-105.(Wang Ying-bo,Huang Qi-tao,Zheng Shu-tao.Dynamic modeling and analysis of a parallel manipulator using Simulink and Sim Mechanics[J].Journal of Harbin Enginering University,2012(1):100-105.)