基于改进牛顿迭代法的混联机构正解分析  

Forward Kinematics Analysis of Hybrid Mechanism Based on Improved Newton Iteration Method

作　　者：宋孝宗[1] 王笑荣 付海涛 赵慧龙 SONG Xiaozong;WANG Xiaorong;FU Haitao;ZHAO Huilong(School of Mechanical Engineering,Lanzhou University of Technology,Lanzhou 730050,China)

机构地区：[1]兰州理工大学机电工程学院,甘肃兰州730050

出　　处：《机械制造与自动化》2023年第4期16-19,共4页Machine Building & Automation

基　　金：国家自然科学基金资助项目(51565031)。

摘　　要：为解决XY-3-RPS混联机构运动学正解求解困难、求解效率低等问题,提出一种利用RBF神经网络改进牛顿迭代的算法。运用闭环矢量法建立混联机构的正向运动学方程。在混联机构的运动学逆解中选取适量的训练样本,通过RBF神经网络进行训练,将训练后的样本估计值作为迭代初值,进行运动学正解的迭代。与牛顿迭代算法的结果相对比,该算法具有更高的精度和更快的迭代速度。To overcome the difficulty and low efficiency in solving forward kinematics of XY-3-RPS hybrid mechanism,a Newton iteration method improved by radial basis function neural network was proposed.By closed-loop vector metho,the forward kinematics equations of the hybrid mechanism were established.According to the inverse kinematics solution of the XY-3-RPS hybrid mechanism,appropriate training samples were selected and trained through RBF neural network.The estimated value of the sample after training was taken as the initial value of Newton iteration method to carry out the solution of the forward kinematics.Compared with the result of Newton iteration method,the improved algorithm has higher accuracy and faster iteration speed.

关 键 词：混联机构 正向运动学 RBF神经网络 牛顿迭代法 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]