采用双RBF神经网络控制的机械手末端位姿研究

Research on the end pose of manipulator controlled by double RBF neural network

为了提高并联机械手末端位姿定位精度和数控机床加工精度,采用双径向基函数(RBF)神经网络控制器,并对机械手定位误差进行仿真验证。给出平面并联机械手运动装置简图,分析并联机械手末端运动位置,推导出执行器运动速度和加速度方程式。设计并联机械手双RBF神经网络控制器,采用Matlab软件对并联机械手运动位姿进行仿真,并与单RBF神经网络控制器进行对比。结果显示:在单RBF神经网络控制器作用下,并联机械手横向位移、纵向位移和平台角位移跟踪误差较大;在双RBF神经网络控制器作用下,并联机械手横向位移、纵向位移和平台角位移跟踪误差较小。采用双RBF神经网络控制器,可以提高并联机械手末端位姿定位精度,从而提高数控机床的加工精度。

In order to improve the positioning accuracy of the end position and posture of the parallel manipulator,and thus improve the machining accuracy of the computer numerical control(CNC)machine tool,a double radial basis function(RBF)neural network controller is adopted,and the positioning error of the manipulator is simulated and verified.The schematic diagram of the planar parallel manipulator is given,the end motion position of the parallel manipulator is analyzed,and the motion velocity and acceleration equations of the actuator are derived.The double RBF neural network controller of the parallel manipulator is designed.The kinematics and posture of the parallel manipulator are simulated by Matlab software,and compared with the single RBF neural network controller.The results show that under the action of single RBF neural network controller,the tracking errors of lateral displacement,longitudinal displacement and platform angular displacement of parallel manipulator are large;Under the action of double RBF neural network controller,the tracking errors of lateral displacement,longitudinal displacement and platform angular displacement of parallel manipulator are small.The double RBF neural network controller can improve the positioning accuracy of the end position of the parallel manipulator,and thus improve the machining accuracy of the CNC machine tool.