Can machine learning revolutionize manufacturing scheduling? This paper explores a dynamic scheduling approach for manufacturing systems, leveraging the power of machine learning to overcome the limitations of traditional dispatching rules. The research addresses the challenge that dispatching rules often perform inconsistently depending on the system's current state, highlighting the need for a more adaptive and intelligent scheduling strategy. Machine learning is applied to select the most appropriate rule at each moment, optimizing overall system performance. The methodology involves five key steps: defining control attributes, creating training examples, acquiring heuristic rules via machine learning, selecting dispatching rules based on these heuristics, and testing the approach's performance. The heuristic rules are obtained using a machine learning program, by considering control attributes that identify manufacturing patterns. The proposed approach is applied to both a flow shop system and a classic job shop configuration, demonstrating its versatility. The results show a significant improvement in system performance compared to traditional dispatching rules. This research demonstrates machine learning's potential to enhance manufacturing efficiency and provides a framework for developing more intelligent and adaptive scheduling systems.
Published in the International Journal of Foundations of Computer Science, this paper aligns perfectly with the journal's focus on theoretical computer science and its applications. The research contributes to the field by exploring the use of machine learning techniques for optimizing complex systems, specifically in manufacturing. The paper's emphasis on algorithmic approaches and performance analysis is highly relevant to the journal's scope.