摘要
用ProE软件建立了水轮机修复专用机器人的本体模型,在此基础上,利用Matlab循环程序求出机器人在给定关节变量下的运动学正解,以此作为训练样本,通过逐次训练6输入、6输出、2个隐含层的BP神经网络,得到机器人从工作空间到关节变量空间的非线性映射,从而实现了水轮机修复专用机器人运动学逆解的计算。
The noumenal model of special purpose robot for repairing hydro-turbine was established by means of Pro-E software. Taking this as a base and using Matlab circulatory program the kinematics forward solution of robot under given joint-variables was found. Taking this as a sample and by way of progressive training on BP neural network with six inputs, six outputs and two hidden layers, nonlinear mapping of the robot from working space to joint-variable space was obtained, thus achieved the calculation of the kinematics inverse solution for the special purpose robot used for repairing hydro-turbine.
出处
《机械设计》
CSCD
北大核心
2005年第6期36-38,共3页
Journal of Machine Design
基金
甘肃省中青年自然基金资助项目(19907)
关键词
水轮机修复专用机器人
运动学逆解
BP神经网络
误差补偿
special purpose robot for repairing hydro-turbine
kinematics inverse solution
BP neural network
error compensation