Adaptive robust control based on RBF neural networks for duct cleaning robot
Adaptive robust control based on RBF neural networks for duct cleaning robot
复制标题
DOI：
10.1007/s12555-012-0447-9
复制
发表时间：
2015-02
期刊：
影响因子：
--
通讯作者：
Bu Dexu;Sun Wei;Hongshan Yu;Wang Cong;Zhang Hui
中科院分区：
文献类型：
--
作者：
Bu Dexu;Sun Wei;Hongshan Yu;Wang Cong;Zhang Hui
本文结合神经网络技术和先进的自适应鲁棒理论的优点，提出了一种在存在不确定性和各种扰动的情况下管道清洗机器人的控制策略。首先，介绍了管道清洗机器人的结构，并在实际管道清洗机器人的基础上建立了机器人的动力学模型。其次，利用径向基函数神经网络强大的逼近任意非线性函数到任意精度的能力，对非结构和动态不确定性进行辨识。利用神经网络的学习能力，设计的控制器可以有效地协调控制管道清洗机器人的移动装置和清扫臂的不同动力学。神经网络权值只需在线调整，无需冗长的离线学习。在此基础上，提出了一种基于RBF神经网络的自适应鲁棒控制方案，即使在存在外部扰动和不确定性的情况下，也能保证轨迹的准确跟踪。最后，基于Lyapunov稳定性理论，严格保证了整个闭环系统的稳定性和跟踪误差的一致最终有界性。仿真和实验结果表明，所提出的控制方法能够保证整个系统以良好的性能收敛到期望的流形。
In this paper, a control strategy for duct cleaning robot in the presence of uncertainties and various disturbances is proposed which combines the advantages of neural network technique and advanced adaptive robust theory. First of all, the configuration of the duct cleaning robot is introduced and the dynamic model is obtained based on the practical duct cleaning robot. Second, the RBF neural network is used to identify the unstructured and dynamic uncertainties due to its strong ability to approximate any nonlinear function to arbitrary accuracy. Using the learning ability of neural network, the designed controller can coordinately control the mobile plant and cleaning arm of duct cleaning robot with different dynamics efficiently. The neural network weights are only tuned on-line without tedious and lengthy off-line learning. Then, an adaptive robust control scheme based on RBF neural network is proposed, which ensures that the trajectories are accurately tracked even in the presence of external disturbances and uncertainties. Finally, based on the Lyapunov stability theory, the stability of the whole closed-loop control system, and the uniformly ultimately boundedness of the tracking errors are all strictly guaranteed. Moreover, simulation and experiment results are given to demonstrate that the proposed control approach can guarantee the whole system converges to desired manifold with well performance.