聚类回归分析在滚齿机热误差建模中的应用

Application of Clustering Regression Analysis to Thermal Error Modeling of Gear Hobbing Machine

利用聚类回归分析方法的基本原理,研究了温度传感器在滚齿机上的优化布置策略,并将温度测点从原先的11个减少到4个,完成了温度变量的优选.利用优选的温度变量,采用最小二乘法进行回归建模,得到热误差模型,并利用该模型在Y3150K型滚齿机上进行热误差补偿实验.结果表明,该建模方法不但增强了热误差建模的鲁棒性,提高了齿轮加工精度,而且节省了工作量与成本.

Based on the culstering regression analysis theory, this paper studied the optimization of temperature sensor positions on gear hobbing machine, the thermal sources were cut down from 11 to 4 and the optimized selection of temperature variables was fulfilled. According to least squares theory, the four selected measurement points were used for regression analysis and the thermal error model was presented. Finally, the compensation model was tested using Y3150K hobbing machine. The result shows that the method not only improve the gear machining accuracy and the robustness of thermal error modeling but also save the time and cost.