基于GA-BP网络的多轴机床热误差建模

Modeling of the Multi-axis NC Machine Tool Thermal Error Based on GA-BP Neural Network

为最大限度减少热误差对多轴联动机床加工精度的影响,综合遗传算法全局收敛性和人工神经网络局部搜索快速性的优点,提出一种基于遗传算法优化BP网络隐层节点数及初始值的机床热误差建模方法。运用Matlab-GUI工具开发了具有通用性的交互式多轴机床热误差建模仿真系统,通过与传统的BP神经网络进行对比分析及试验论证,证明该模型预测精度更高、通用性强。

In order to eliminate the influence of thermal error on the machining precision, a novel model for thermal error in machine tools is proposed by using the BP via GA to optimize BP neural network structure and parameters. A versatile interactive thermal error modeling simulation system interface is designed by using Matlab-GUI, the re- sults show that the present model performs better than the traditional BP model. It is good in fitting and precise in prediction for a great number of working conditions.