变工况下基于迁移学习融合域内对齐的机床主轴热误差模型

热误差建模和补偿是提高机床加工精度的重要手段。 将得到的热误差模型应用到类似或相近任务中,对减少模型构建和数据收集的成本具有重要意义。 本文提出了一种简易迁移学习(EasyTL)融合域内对齐的主轴热误差建模方法,以实现不同工况下误差模型的迁移复用。 建立基于域内对齐和距离矩阵全组合择优的热误差迁移模型参数选取方法,获得最优组合。 进一步分析不同类型的域内对齐和距离矩阵各自对模型迁移性能的影响。 最后,将迁移模型与 kNN 典型机器学习模型和卷积神经网络深度模型进行比较验证,分别预测不同工况下主轴 Z 向和 Y 向的热误差。 此外,根据预测的主轴热误差进行工件补偿加工实验。 该方法为热误差建模及补偿提供了一种新思路。

Thermal error model of machine tool spindle based on in-domain alignment and transfer learning under variable working conditions

Thermal error modeling and compensation is an important tool to improve the machining accuracy of machine tools. It isimportant to apply the obtained thermal error models to similar tasks to reduce the cost of model construction and data collection. In thisarticle, an easy transfer learning (EasyTL) with intra-domain alignment method for spindle thermal error modeling is proposed to realizethe transfer reuse of error models under different working conditions. Further, the respective effects of different types of intra-domainalignment and distance matrices on model migration performance are analyzed. Finally, the EasyTL model is compared and validated withmachine learning kNN and deep learning CNN to predict the thermal errors of the Z-direction and Y-direction of spindle under differentworking conditions, respectively. This method provides a new idea for modeling and compensating the thermal errors of machinespindles. In addition, a workpiece compensation machining experiment is carried out according to the thermal error of the spindleestablished by the thermal error prediction. The average error of the workpiece after compensation is reduced. This method provides anew idea for the thermal error modeling and compensation.