System-level multi-objective design optimization of machine tools

Design optimization of machine tools is a crucial step to achieve improved dynamic performance, accuracy, and energy efficiency. Most existing methods focus on component-level optimization which neglects the critical effects of simultaneous modification of the machine tool components on design space exploration. This study presents a comprehensive approach for system-level multi-objective optimization of machine tools. Fundamental aspects of the machine tool including thermal and dynamic stability, stiffness, and fatigue life are included in optimization to ensure the optimized design maintains or improves the performance of the base design while mass reduction is achieved. A machine learning model is trained as a surrogate for finite element analysis to guide the optimization process more efficiently. A design parameterization method is proposed to achieve substantial design modifications across the machine tool using a minimum number of design variables to further increase computational efficiency. An energy-based reduced-order representation for spindle nose frequency response functions (FRFs) is introduced to enable computationally feasible prediction of dynamic stability indicators without requiring complete FRF curves. It is demonstrated that the proposed design optimization approach achieves up to 20% mass reduction in machine tool components without sacrificing machine performance.