粒子群算法求解具有机器灵活性的FFSP

Particle swarm optimization for flexible flow shop scheduling problem with machine flexibility

对柔性流水车间调度问题(FFSP)进行了分析阐述,在此基础上对某饲料厂的饲料生产过程建立了具有机器灵活性的柔性流水车间调度模型,该模型中存在多台制粒机,既能加工大颗粒饲料,又能加工小颗粒饲料,但是必须在开始加工之前确定各台机器的用途,增加了柔性流水车间调度的难度。利用新型的粒子群算法以最小化最大完工时间为目标对该模型求解,为了克服粒子群算法易陷入局部极值的缺点,提出基于位置相似度的邻域结构,并对邻域内的较优粒子采用基于最大完工时间排序的学习方式进行局部搜索。实验结果表明,该方法有利于克服粒子群算法的早熟缺陷,有效地解决了饲料生产调度问题,有一定的应用价值。

Flexible flow shop scheduling problem （FFSP） is analyzed in this paper. Then a FFSP model with machine flexibility is established for the feed production process of a feed factory. In the model there are many sets of granulator, which can product not only small particles but also big particles. But the use of the machine must be determined before the start of the processing, which increased the difficulty of the flexible flow shop scheduling problem. Using new particle swarm optimization （PSO） algorithm to solving the model for the goal of minimize the maximum completion time. In order to overcome the shortcoming of easily trapped in local minima of pso, we put forward a neighborhood structure based on position similarity, and take a makespan rank based learning method for local search. The experimental results show that this method is beneficial to overcome the premature defects of pso algorithm, and effectively solve the problem of feed production scheduling. It has certain application value.