This thesis explores the integration of production and maintenance operations in industrial planning, addressing three key challenges: misalignment between production and maintenance schedules, the need for real-time maintenance execution, and the gap between academic research and practical application. It proposes a comprehensive framework that synchronizes production, maintenance, and resource scheduling across operational and execution layers.nnThe first part introduces a hybrid genetic algorithm to optimize scheduling for unrelated parallel machines, considering constraints like setup times and limited personnel. This integrated approach significantly improves production efficiency and reduces tardiness, as shown in a semiconductor manufacturing case study.nnThe second part focuses on real-time maintenance execution. Traditional static planning is replaced with dynamic methods using digital twins and deep reinforcement learning, which adapt to live data such as machine states and product quality. These methods were tested in single and multi-line environments, demonstrating enhanced throughput and reduced disruptions.nnFinally, the framework was implemented in collaboration with Nexperia, integrating real-time maintenance planning into existing IT systems. Pilot trials showed a 0.7% increase in production time, translating to over a million dollars in savings. Overall, the thesis bridges theory and practice, offering a scalable solution for modern manufacturing and laying the groundwork for future research in complex, data-driven environments.