基于FCM聚类和RBF神经网络的机床热误差补偿建模

Thermal Error Compensation Modeling Based on Fuzzy C-means Clustering Algorithmand RBF Neural Network Modeling

热关键点的选择和热误差建模技术是决定热误差补偿是否有效的关键,对提高数控机床的加工精度至关重要。为了实现对数控机床热误差的补偿控制,文章利用模糊C均值(FCM)聚类方法,对机床上布置的温度测点进行优化筛选,将温度变量从20个减少到4个,然后给出了基于RBF热误差补偿建模方法。通过建模实例表明,文章提出的建模方法,在保证补偿模型精度的同时有效减少了温度测点,降低了变量耦合影响,并提高了补偿模型的鲁棒性。

The selection of thermal critical points and thermal error compensation modeling technique are crucial in deciding the effectiveness of thermal error compensation and important for improving machining accuracy of numerical control（NC） machine.In order to realize the compensation of the thermal error of NC machine,the temperature measurement points are optimized based on the fuzzy C-means（FCM） clustering algorithm and the number of temperature measurement points is cut down from 20 to 4,then the thermal error compensation model is established based on RBF neural network.The experiment result shows that the modeling method is proposed by this paper not only ensure the precision of the model,reduce the measurement points and avoid the correlation of the measurement points,but also improve the robustness of thermal error modeling.