Vishal Sharma
Engineering
September 2025
The paper examines the use of Artificial Intelligence (AI) approaches in optimising production planning and scheduling in manufacturing systems and limitations of conventional approaches in complexity, volatility, and real-time disruptions. Five AI algorithms i.e. Genetic Algorithms, Reinforcement Learning, Neural Networks, Ant Colony Optimisation and Deep Q-Learning were built leveraging benchmark datasets and simulated-generated data, and compared with standard scheduling strategies. The performance assessment was conducted on the major aspects such as the make span, lead time, machinery utilization, cost of operation, flexibility and capacity to absorb disruptions. The outcome indicates that the AI-based approaches are highly efficient compared to the classical scheduling and Deep Q-Learning method demonstrates the most efficient outcome, flexibility, and recovery time under a rigorous response to meeting demands but needs more training time. This balance of performance versus computational requirement made Ant Colony Optimisation and Reinforcement Learning well suited to being used in an industry setting. The results point out the potential of AI to achieve strategic advantages in process efficiency, responsiveness, and reliability, especially in dynamic manufacturing processes.
597- 611
© Airo Journals 2021
