Thermal error modeling for CNC machine tools based on parallel locally weighted linear regression

To shorten the modeling time of thermal errors in mass data, a parallel locally weighted linear regression model is proposed on the basis of locally weighted linear regression and cloud computing platform. The massive data is divided into multiple data sub-blocks and the data of each sub-block is analyzed and processed by the cloud platform simultaneously, followed by the merging of results. Experiments are carried out on a CNC gear grinding machine, the results of which show that the proposed method performs modeling on the basis of mass data in a short period of time. In addition, the model enjoys high precision and strong robustness. This method provides reference for the application of big data in thermal error prediction of CNC machine tools.