Design and multi-objective optimızation of a high-speed spindle considering dynamic and thermal behaviors

High-speed spindles are critical machine tool components with complicated underlying behaviors. Among these, the dynamic and thermal behaviors are the most dominant and interdependent, exhibiting strong links to multiple spindle parameters and performance indicators. This creates a complex design space for spindle designers, warranting the need for a comprehensive optimization approach to maximize spindle performance. While this optimization avenue has been investigated before, the full scope of this problem remains unexplored, particularly in its application to integrated thermal-dynamic models.This thesis presents a multiobjective optimization approach targeting various facets of thermal and dynamic behaviors of high-speed spindles. A novel optimization approach is developed based on the Teaching Learning Based Algorithm (TLBO) and Non-Dominated Sorting Algorithm (NSGA-III) to identify optimal spindle design configurations and reveal the inherent tradeoffs between dynamic and thermal behaviors of spindles. The optimization study was implemented as a step-by-step procedure across different stages of the spindle design process, considering the prevalent objectives, constraints, and design status at each stage. This approach towards multiobjective optimization can serve as a practical framework for optimizing complex mechanical systems. Additionally, a detailed commentary is also included on the practical design and manufacturing constraints regarding high-speed spindles.A proof-of-concept study for the reverse identification of spindle parameters using modal parameters is also presented. A database-assisted machine learning approach based on the XGBoost algorithm is used for this. High prediction accuracies were obtained for the machine learning system, indicating that such studies can be explored further.