Study of Evolutionary Algorithms for Solving Facility Layout Problems

在许多工厂或工厂中，设施的位置和设施之间的配置关系到材料处理的有效性和其他操作问题。在本研究中，设施空间位置的确定问题称为设施布局问题。在本研究项目中，我们将重点放在运营和其他要求以及物资处理上，并考虑了设施之间的邻接偏好。通过使用邻接优先选项，可以详细考虑更好的设施配置。然而，由于邻接偏好项的满意度被表示为二元变量的离散值0或1，因此问题较为复杂。根据问题的大小，我们将布局问题分为几类。在设施数小于或等于14的情况下，将该问题视为小规模问题，提出了一种基于分枝定界法求解此类问题的新算法。在Num…的情况下在BER大于14的情况下，将问题视为大规模问题，提出了基于模拟退火法、遗传算法和禁忌搜索方法的进化算法。通过数值算例对这些算法进行了实验比较，得到了搜索的有效性和问题的数据结构。提出了一种将模拟退火法与原邻域搜索法相结合的新算法。通过在最近的个人计算机上使用我们的算法，对于在实际计算时间内不可能找到最优或半最优解的大型问题，能够找到最优解或最优解。针对多层设施布局问题，在问题较小的情况下，提出了一种基于分枝定界法的算法，并考虑了对称布局等基于模拟退火法的算法来解决较大问题。较少

In many plants or factories, the location of facilities and the configuration between facilities are related to the effectiveness of material handling and other operational issues. In this study, the problem of deciding spatial location of facilities is called as facility layout problem. In this research project, we focused on operational and other requirements as well as material handling, and considered adjacency preference between facilities. By using adjacency preference, preferable configuration of facilities can be considered minutely. However the problem is complicated because the satisfaction of adjacency preference items is represented to 0 or 1 discrete value of binary variables. We classified layout problems into some types with the problem size. In the case of the number of facilities less than or equal to 14, the problem is regarded as the small size problem, and we developed a new algorithm based on branch and bound method for solving such problems. In the case of the num … More ber is greater than 14, the problem is regarded as large size one, we developed evolutionary algorithms based on simulated annealing technique, genetic algorithms and tabu search method. We compared those algorithms by experiments using numerical examples, and got information of effectiveness of searching and data structure of problems. Moreover, far solving very large scale problem of single floor facility layout; a new algorithm combined simulated annealing and the original neighborhood search method is developed. By using our algorithm on recent personal computers, for large size problem that it is not possible to find the optimum or semi-optimum in practical computation time, it became to be able to find the optimum or the excellent solution. For solving multi floor facility layout problems, we proposed an algorithm based on branch and bound method in the case of small problem, and considering of symmetry layout and other algorithm based on simulated annealing technique to solve large problems. Less