Statistical model comparison based on variation parameters for monitoring thermal deformation of workpiece in end-milling

In order to guarantee the accuracy of a workpiece’s dimensions in end-milling operation, the thermal deformation needs to be predicted in a timely manner during the end-milling process. Thermal deformation, which is generated by cutting heat, is hard to measure during the end-milling process, but the local temperature of the workpiece can be conveniently and accurately measured by a thermocouple. We have thus developed an efficient and accurate model to describe the relationship between the local temperature and thermal deformation of a workpiece for monitoring the workpiece’s deformations during the end-milling process. We systematically conducted a finite element method (FEM)–based numerical simulation of a workpiece during end-milling and then scientifically validated it with a previous machining experiment (Human Friendly Mechatr 389-394, 2001) under the same boundary conditions. Currently, the process models for predicting the thermal deformation of end-milling is commonly utilized before actual machining. In order to establish a timely, dynamic, and efficient empirical model for use in the various scenarios related totally in-process of end-milling machining, we have developed a statistics-based temperature monitoring point selection method and a statistical model utilizing multiple linear regression (MLR), the Akaike information criterion (AIC), and a p value index. FEM-based thermal simulation during the end-milling process is utilized in the statistical model, which correlates the deformation at the machining point with the temperature at the measuring points (Yang, 2021). However, the applicability of this model in various machining situations has not been sufficiently evaluated. This article deals with the comparison of statistical modes to adapt the model in various machining situations. After demonstrating the accuracy of the proposed statistical model under various boundary conditions during the end-milling process, we propose two modified statistical models, namely, a modified coefficient statistical model (MCSM) and an adjusted statistical model (ASM), to more appropriately express the relationship between the temperature at the monitoring points and the thermal deformation at the machining point of the workpiece surface. Error analysis of the two models indicated that both provided high accuracy for describing this relationship. In particular, the MCMS improved in the case of 80% evaluation. In the case where only the ranges of the machining parameters are known in advance, the MCSM is preferred for describing the relationship between the temperature of the monitoring points and the thermal deformation at the machining point.