柔性关节机操手的神经网络控制

本文在关节柔性较弱的情况下，对柔性关节机器人操作手的轨迹跟踪问题，提出了一种基于奇异摄动理论的机器人神经网络控制设计方法，在一般框架下证明了系统跟踪误差最终一致有界，并且可以通过选取增益矩阵使该误差界任意地小. 该方法克服了对模型参数线性化条件的要求，无需求解回归矩阵，因而具有很强的鲁棒性和模型推广能力. 数值试验表明，所提出的控制方法是可行且有效的.

Neural Network Control of Flexible-joint Robot Manipulators

In this paper, for flexible-joint robot manipulators with weak flexibility, we propose a neural network trajectory-tracking strategy based on singular perturbation theory. Under general assumptions, we prove that the tracking error is ultimately uniformly bounded and that the corresponging ultimate bound can be sufficiently decreased by modifying the feedback gain matrix. Since the linearization assumption of the unknown parameters is removed, the regression matrix need not be conmputed. Therefore, the proposed method has great robustness and the ability of model generalization. The numerical simulation shows that the proposed method is feasible and efficient.