基于有序聚类的机床热误差测点优化

Optimizing of CNC machine tool temperature measuring points based on sequential cluster method

温度测点的选择直接影响数控机床热误差补偿模型的性能。考虑到温度有序传递的特点,提出了有序聚类测点优化的方法。以试验数据为基础,计算类直径并比较目标误差函数;然后对温度变量分类,确定最佳分类数;通过计算热误差和温度之间的相关系数,确定最优测点。采用定位误差分解建模法结合选取的最优测点建立热误差预测模型,分别与模糊聚类和变量分组测点优化建立的模型进行比较,试验结果表明,有序聚类测点优化法精度较高,具有一定的应用前景。

The performance of CNC machine tool thermal error compensation model is affected directly by the selection of temper- ature sensor placement. An optimizing method of sensor selected by sequential cluster is proposed, given that the characteristic temperature is transferred in order. Based on experiment data, classification diameters are calculated and their objective function are compared. Then the optimizing classification number and optimizing sensor placement are determined by the calculation of correlation coefficient between thermal error and temperature value. Thermal error prediction model is established with a position error modelling method and the selected points. In comparison with models established by fuzzy cluster and variable grouping, the results show the proposed method can accurately predict and has potential application prospect.