摘要
从优化网络结构角度出发 ,对快速BP网络用自构方法进行了改进 ,克服了以往只能依靠实验结果选择合适隐节点个数的局限性 ,使神经网络隐节点个数的选取更加合理 .在此基础上 ,提出了一种基于自构形快速BP网络的并联机器人位置正解方法 ,并以新型 6 HURU并联机器人为例进行了求解 .结果表明 ,位置正解的平均误差可小于 0 .15mm和 0 .0 6° ,能够满足一般应用的精度要求 。
To optimize the structure of neural network, this paper improves the quick BP network by self configuration method, which overcomes the restrictions of selecting the number of suitable hidden nodes only through experiments in the past and makes it more reasonable to select the number of hidden nodes of neural network. On this basis, this paper proposes a method for forward displacement solution of parallel robot based on self configuration quick BP neural network. As an example, the forward displacement solution of novel 6 HURU parallel robot is resolved, and the results show that its average errors are less than 0.15mm and 0.06°, and meet the precision requirement of general applications. The validity and feasibility of the proposed method are proved.
出处
《机器人》
EI
CSCD
北大核心
2004年第4期314-319,共6页
Robot
关键词
并联机器人
位置正解
快速BP网络
自构
parallel robot
forward displacement solution
quick BP network
self configuration
