BP神经网络求6PTRT型并联机器人正运动学精确解

The BP neural network for 6PTRT type parallel robot forward kinematics precise solution

采用BP神经网络,利用位置逆解结果,通过训练学习,实现操作从关节变量空间到工作变量空间的非线性映射,从而求出6PTRT型并联机器人的正运动学解。计算实例表明单用BP神经网络得到的精度并不高,所以为提高正解结果精度,引入误差补偿算法,并设计相应软件,所得数据表明,该算法计算精度高。

By using BP neural network,by position inverse results,through training study,its operation from joint variable space to work variable space nonlinear mapping,thus find out 6PTRT type of parallel robots forward kinematics solutions.Calculation sample showed designates the BP neural network obtained the accuracy is not high,so in order to improve the precision of positive solution results,the introduction of error compensation algorithm,and appropriate design software,income data show that the algorithm is high precision.