Artificial Intelligence is everywhere in the conversation about manufacturing and still surprisingly rare on the actual factory floor. The gap between what AI can theoretically do for operations and what companies have actually managed to implement is one of the most consequential disconnects in modern industrial management. Delivery dates are still missed. Machines still fail without warning.

The tools to fix these problems exist. Machine learning algorithms that predict due date adherence before a single component has been touched. Artificial neural networks that detect the early signatures of machine failure days before a breakdown occurs, eliminating scrap, cutting costs and extending equipment life. Models that identify where a production line will choke next; not after it happens, but in time to do something about it. The academic evidence for each of these applications is growing. What is missing is a coherent map of the territory and a rigorous test of what actually works when the rubber meets the road.

This thesis builds that map and then takes it further. Starting from a structured literature review of AI and machine learning applications across the Operations Management domain (production planning and control, predictive maintenance, bottleneck detection and beyond), the research identifies the most promising methodologies and puts them to the test, either in a simulated environment built on Digital Twin technology or in a real industrial case study.

The output is not just an academic contribution. It is a methodology that a company could pick up and implement, grounded in evidence, tested under realistic conditions and designed to close the gap between what AI promises and what operations managers can actually use.

If you want to work at the precise intersection where data science meets the shop floor and prove that machine learning can do something real for the companies that need it most, this is the thesis for you.