机器人末端臂轨迹跟踪自适应控制设计

本文提出了新颖的机器人末端臂轨迹跟踪自适应控制方法。该方法与已有的神经网络模型不同之处在于数据首先利用运动学反解求出机器人各关节旋转的角度,然后应用径向基函数自组织进行神经网络学习生成模糊规则,利用监督学习算法(SLA)、最小二乘法(LMS)、反向传播算法(BP)和聚类分析的方法在线优化控制规则以及隶属函数的参数。仿真结果表明,该方法不但规则生成的时间少,有效的防止了规则数爆炸,而且在机器人轨迹跟踪控制的应用中效果好。

Design on adaptation control for robot end-effector trajectory tracking

This paper presents a novel adaptation control method for trajectory tracking of robotic end-effector. The method is different from presented neural network.First, datum is transformed to revolve angular using by kinematics inverse algorithm, and then self-or-ganizing generate control rules by radial basis function neural networks. Optical rules and parameters of membership function are up-dated by the supervised learning,the least mean square algorithm,the back-propagation algorithm and clustering analytic approach. The simulation results demonstrate the performance of the proposed method not only reduce the time of generation rules, avoid explod-ing the number of rule, but also the excellent tracking performance can be achieved in simulation.