• 论文 •    

基于神经网络的并联机床结构精度的研究

魏永庚,王知行   

  1. 哈尔滨工业大学机电学院,黑龙江哈尔滨150001
  • 出版日期:2003-09-15 发布日期:2003-09-25

Structure Precision for Parallel Machine Tool Based on Neural Network

WEI Yong-geng,WANG Zhi-xing   

  1. Sch.of Mechanical and Electrical Eng., Harbin Inst.of Tech., Harbin150001, China
  • Online:2003-09-15 Published:2003-09-25

摘要: 针对并联机床各运动副误差难以直接测量的问题,设计了基于神经网络的并联机床结构精度标定测试系统。利用神经网络建立并联机床误差补偿的数学模型,将非线性误差作线性处理,修正难以模型化因素造成的误差;模型化因素造成的误差由三坐标测量机修正。实验证明,误差的90%可由三坐标测量机修正,而残留误差的1/3~1/2可用神经网络修正。

关键词: 神经网络, 并联机床, 三坐标测量机, 精度

Abstract: Because the kinematic pair error of Parallel Machine Tool is difficult to be measured directly, this paper designs the measuring system on structure precision for Parallel Machine Tool based on neural network. The maths model based on neural network is established to compensate Parallel Machine Tool errors, and turns nonlinear errors into linear errors to modify the errors which are difficult to establish maths model. Three-coordinate Measure Machine modifies the errors that are easy to establish maths model. The test result indicates that Three-coordinate Measure Machine can modify 90 percent of errors and neural network can modify 1/3~1/2 of surplus errors.

Key words: neural network, parallel machine tool, three-coordinate measure machine, precision

中图分类号: