Thermal error modelling for a CNC machine based on data driven analysis

CNC machines tools are affected by precision error and the main cause of it, for about the 70 percent, is the thermal error. This thermal deformation compensation is becoming the main target for research in theory and industry environment. The internship executed in Giuliani, a worldwide known producer of machines for the lock industry, has the goal to study and create an efficient model for the units of the TDRILL, the company’s rotary transfer machine. The focus is on the generation of thermal error’s model based on data driven analysis; in particular two approaches are been followed: the first one consists in the development of a multiple linear regression model based on the least squares criteria while the other one uses the neural networks. The project mixed theoretical concepts for the initial design and practical consideration and solution for the improvement of the model referring to the specific machine. The final model’s residuals are under the accuracy request by the company (10 um) in all the different simulations that are been executed during the internship.