基于有限元及贝叶斯网络的数控滚齿机热误差建模研究

Thermal error modeling of CNC gear hobbing machine based on finite element and Bayesian network

为减少热误差对数控滚齿机加工精度的影响,进行数控滚齿机床的热-结构仿真分析,基于贝叶斯网络理论建立了数控滚齿机热误差模型。利用有限元软件模拟数控滚齿机加工时的温升和热位移情况;基于仿真数据,采用模糊聚类对初始选择的多个温度变量进行分析,减少了温度变量间的耦合,再比较主轴热位移与温度变量的相关性,对聚类结果进行筛选得到建模所需的热关键点,减少了热误差实验测温的传感器数量。实验采集滚齿机的热关键点的温升和滚刀主轴热误差数据,对提出的热误差预测模型与实验测量数据进行比较,两者最大误差为3.5μm,表明该模型可有效预测滚齿机的热误差,为机床的热误差补偿提供了理论依据。

To reduce the impact of thermal errors on the machining accuracy of CNC gear hobbing machines processing, the thermal-structural simulation of CNC hobbing machine is analyzed and the thermal error model of CNC hobbing machine based on bsayesian network is established. The rise of temperature and thermal displacement of the CNC hobbing machine are simulatedby using finite element software.Based on the data from simulation, Fuzzy clustering is used to analyze the initially selected temperature variables to reduce the coupling between temperature variables, the correlation between the thermal displacement of the main axis and the temperature variables is compared, and the clustering results are screened to obtain the thermal key points required for modeling, which reduces the number of sensors for temperature measurement in thermal error experiments.The rise of the thermal key points’ temperature of the gear hobbing machine and the thermal error data of the hob spindle are obtained experimentally. The proposed model of thermal error prediction is compared with the data of the experiment, and the maximum error is 3.5μm, which indicates that the model can predict the thermal error of the machine effectively and provide a theoretical basis for the compensationof thermal error of the machine tool.