运用多体理论和神经网络的机床热误差补偿

利用多体系统理论,在典型体的坐标变换中,加入了位移误差矢量和位置误差矢量,形成了具有普遍意义的坐标变换,根据机床拓扑结构的低序体阵列,建立了机床通用误差计算模型。同时,对机床的主轴热变形和床身热变形进行了建模和辨识,通过5个温度敏感点的监测,用常规的5点法对机床主轴热变形进行研究,运用神经网络方法(RBF)建立温度与变形参数模型,将误差参数集成到通用误差模型中。在Makino四轴加工中心进行试验研究,设计出一套多个凸台的空间曲面,比较了不同凸台上的4个点补偿前后空间轮廓数据,误差减少60%,补偿效果显著。

Thermal Error Model and Compensation of Machining Tool Using Multibody System Theory and Neural Network

Abstract Through analysis of the basic transformation of the typical body, the error transformation of the position vector and the displacement vector between two adjacent bodies are proposed. By means of the lower numbered body array, which is constructed by the topological construction of a system, a complicated model including thermal errors is presented for four axes machining centers based on the Multi-body System Theory. The spindle thermal errors are measured by normal five points method. The structure thermal errors of machining centers are determined by using a RBF neural network method, which are related with five temperature sensors All the thermal errors are integrated in a model. Meanwhile, a special work piece including many dummy clubs is designed for the error compensation experiment, which is carried out in Makino four-axes machining center. Compared with four points in different dummy clubs on the conditions of the theoretical result , with and without compensation results , the positioning errors are decreased 60%, the accuracy of the work piece is improved obviously.