混合粒子群算法求解多目标柔性作业车间调调度度问题

Hybrid particle-swarm optimization for multi-objective flexible job-shop scheduling problem

柔性作业车间调度问题是生产管理领域和组合优化领域的重要分支.本文提出一种基于Pareto支配的混合粒子群优化算法求解多目标柔性作业车间调度问题.首先采用基于工序排序和机器分配的粒子表达方式,并直接在离散域进行位置更新.其次,提出基于Baldwinian学习策略和模拟退火技术相结合的多目标局部搜索策略,以平衡算法的全局探索能力和局部开发能力.然后引入Pareto支配的概念来比较粒子的优劣性,并采用外部档案保存进化过程中的非支配解.最后用于求解该类问题的经典算例,并与已有算法进行比较,所提算法在收敛性和分布均匀性方面均具有明显优势.

Flexible job-shop scheduling is a very important branch in both fields of production management and com- binatorial optimization. A hybrid particle-swarm optimization algorithm is proposed to study tile mutli-objective flexible job-shop scheduling problem based on Pareto-dominance. First, particles are represented based on job operation and ma- chine assignment, and are updated directly in the discrete domain. Then, a multi-objective local search strategy including Baldwinian learning mechanism and simulated annealing technology is introduced to balance global exploration and local exploitation. Third, Pareto-dominance is applied to compare different solutions, and an external archive is employed to hold and update the obtained non-dominated solutions. Finally, the proposed algorithm is simulated on numerical clas- sical benchmark examples and compared with existing methods. It is shown that the proposed method achieves better performance in both convergence and diversity.