摘要
研究机器人跟踪轨迹控制问题,针对模型未知的机器人系统,为提高跟踪精度和控制性能,提出了一种基于T-S型模糊RBF神经网络的H∞轨迹跟踪控制方法,用模糊神经网络为模型未知的机器人系统建模,克服了系统鲁棒性差,对机动目标跟踪性能差等缺点。然后设计自适应控制器,将H∞控制理论与模糊神经网络有机地结合起来,借助鲁棒补偿项将建模误差及外部干扰衰减到期望的程度以下,而控制器与改进Elman神经网络的结合,便于处理建模有界干扰以及非结构化的未建模的动力学,并进行仿真。仿真结果表明了所提出的控制算法的可行性。
In this paper,a new control strategy is proposed for robot manipulators with uncertainties.The control scheme combines H∞ control theory with Takagi-Sugeno（T-S） model fuzzy Elman neural network algorithm organically,and the influence of both the modeling errors and the external disturbances can be attenuated to a desired level.It overcomes the poor system robustness and poor tracking performance characteristics.The controller adopts fuzzy neural network to estimate robotic dynamics directly without changing it.Combining controller with Elman neural network is easy to process modeling bounded disturbances and unstructured unmodeled dynamics.Based on Lyapunov method,learning adaptive law is given and H∞ tracking performance is illustrated.The simulation studies based on a 2-DOF robot verify the effectiveness of the proposed algorithm.
出处
《计算机仿真》
CSCD
北大核心
2010年第8期145-149,共5页
Computer Simulation
