Comprehensive Analysis of Temperature-Sensitive Points Across Machine Tool Structures Using Highly Redundant Temperature Data and Sparse Modeling

Thermal influences account for up to 75% of errors in precision machining, highlighting the critical requirement for effective thermal error compensation. In this study, we employed a large-scale array of temperature sensors interconnected in series together with least absolute shrinkage and selection operator (LASSO) regression to determine the optimal number and placement of temperature sensors for precise thermal error estimation. Temperature data from 307 points were collected under six operational patterns on a three-axis horizontal machining center and subjected to correlation analysis. Distinct correlation map trends emerged for each pattern, underscoring the difficulty of removing highly correlated coefficients. Further, by tuning the LASSO regularization parameter, we reduced the sensor count by 76% while keeping the root mean square error below 10 µm, thereby shifting the priority of sensor locations. These findings demonstrate a practical, data-driven pathway for deploying minimal yet highly informative sensor sets, enabling cost-effective and physically interpretable thermal error compensation in next generation precision machine tools.