A whale optimization algorithm with Q-learning and processing-sequence-based distance for solving distributed blocking multi-product batch plant scheduling with sequence-dependent set-up times

Le résumé fourni par la source

Batch processing is crucial in modern industries such as food processing and pharmaceuticals, with scheduling under blocking constraints posing a challenge. This study addresses distributed blocking batch plant scheduling with sequence-dependent set-up times. A mixed-integer linear programming (MILP) model is developed to represent the problem, and a novel whale optimization algorithm incorporating Q-learning and processing-sequence-based distance (QDWOA) is proposed. Population diversity significantly impacts the performance of evolutionary algorithms, making an effective diversity metric essential. Therefore, this study presents a processing-sequence-based distance (PSD) to quantify individual distances in the population and then defines a diversity metric, percentage population diversity (PPD), to evaluate population diversity. During the encircling prey phase, six neighbourhood strategies are proposed, and Q-learning is used to dynamically select the most appropriate strategy. Experimental results show that the proposed algorithm outperforms several state-of-the-art metaheuristics. Moreover, the MILP problem is solved using CPLEX, yielding optimal solutions for 27 small-scale instances.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.