Reinforcement learning-based alpha-list iterated greedy for production scheduling
Year: 2024
Type: article
Abstract: Metaheuristics can benefit from analyzing patterns and regularities in data to perform more effective searches in \nthe solution space. In line with the emerging trend in the optimization literature, this study introduces the \nReinforcement-learning-based Alpha-List Iterated Greedy (RAIG) algorithm to contribute to the advances in \nmachine learning-based optimization, notably for solving combinatorial problems. RAIG uses an N-List mechanism for solution initialization and its solution improvement procedure is enhanced by Reinforcement Learning \nand an Alpha-List mechanism for more effective searches. A classic engineering optimization problem, the \nPermutation Flowshop Scheduling Problem (PFSP), is considered for numerical experiments to evaluate RAIG’s \nperformance. Highly competitive solutions to the classic scheduling problem are identified, with up to 9% \nimprovement compared to the baseline, when solving large-size instances. Experimental results also show that \nthe RAIG algorithm performs more robustly than the baseline algorithm. Statistical tests confirm that RAIG is \nsuperior and hence can be introduced as a strong benchmark for future studies.
Language: English
Subfield: Industrial and Manufacturing Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $1,500
Funder: Universitetet i Tromsø
OpenAlex ID: W4403329688