A Multiscale Manufacturing Operations Management Study from Discrete-event Dynamics to Machining Dynamics by Machine Learning based Modeling and Optimization

This dissertation advances the fields of manufacturing operations management through machine learning based modeling and optimization across multiple scales. From the scale of factory, a customized genetic programming approach together with adaptive local search is developed to discover effective dispatching rules and generate better customer order sequences for customer order scheduling. From the scale of machine tool, discrete-event dynamics are incorporated into machine shop to formulate a learning-based cost function and optimization models that minimize machine tool costs. From the scale of machining dynamics, a cutting mechanics-based machine learning modeling method is proposed to identify governing equations of machining dynamics by integrating physical knowledge in cutting mechanics with data-driven insights. Taken together, these research studies establish a systematic study linking operations management at different levels, contributing to both theoretical understanding and practical applications in data-driven and intelligent manufacturing systems.