Q-Learning-Based Improved Discrete PSO for Multi-Objective Flexible Job-Shop Scheduling with Worker Fatigue and Dual Resources
提出一种改进离散粒子群算法，通过Q学习自适应调节参数，求解考虑工人疲劳和双资源约束的多目标柔性作业车间调度问题，在最小化完工时间、总成本和工人疲劳不平衡方面优于对比算法。
Flexible Job Shop Scheduling Problem (FJSP) is a critical combinatorial optimization problem in manufacturing systems. This paper presents an Improved Discrete Particle Swarm Optimization (IDPSO) algorithm for the multi-objective flexible job-shop scheduling with worker fatigue and dual resources constraints (FJSP-WF-DRC). This algorithm dynamically classifies particles into three categories: elite particles, common particles and chasing particles. This classification strategy allows the algorithm to apply distinct evolutionary operators tailored to the role of each particle. IDPSO incorporates a Q-learning-based parameter control mechanism that adaptively adjusts key algorithmic parameters and search strategies. Comprehensive experiments were performed to compare IDPSO with several state-of-the-art algorithms including two-stage cooperative discrete differential evolution with Q-learning (QTCDDE), non-dominated sorting genetic algorithm with a multimodal solution preservation mechanism (NSGA-MSPM) and learning-based reference vector memetic algorithm (LRVMA) on various instances. Performance assessment considers three key objectives: minimizing makespan (C<sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">max</sub>), reducing total cost (TC), and minimizing worker fatigue imbalance measured by the Gini coefficient (FGC). Experimental results demonstrate that IDPSO consistently generates superior Pareto fronts.