ASYNCHRONOUS ISLAND MODEL GENETIC ALGORITHM FOR UNIVERSITY COURSE TIMETABLING
ASYNCHRONOUS ISLAND MODEL GENETIC ALGORITHM FOR UNIVERSITY COURSE TIMETABLING
复制标题
大学课程安排的异步岛模型遗传算法
DOI：
--
复制
发表时间：
2014
期刊：
影响因子：
--
通讯作者：
A. A. Gozali
中科院分区：
文献类型：
--
作者：
A. A. Gozali
大学课程编排问题（UCTP）与一般的排课问题类似，但又有一些特殊之处。UCTP涉及到分配讲座活动的时间和房间受到各种硬和软约束。Telkom大学在课程编排方面也有类似的问题。目前的解决方案与信息遗传算法的Telkom大学UCTP仍然存在耗时的问题。本研究采用基于孤岛模型的遗传算法来解决这一问题。本研究的思想是将一个岛屿的局部最优个体与另一个岛屿的局部最优个体进行分布式交换。孤岛模型遗传算法可以在合理的时间内生成高校排课表。这种分布式模型可以比单机模型运行得更快，从而减少约束违反以达到最佳适应度。它可以有更少的约束违反，因为它可以摆脱停滞的局部最优更容易。Isl和模型GA甚至可以为Telkom大学数据集（99.74%）和Purdue数据集（96.80%）产生很高的准确度，用于学生级的排名。
University course timetabling problem (UCTP) is similar to general timetabling problems with some additional unique parts. UCTP involves assigning lecture events to timeslots and rooms subject to a variety of hard and soft constraints. Telkom University has almost similar problem with its course timetabling. The current solution with Informed Genetic Algorithm for Telkom University UCTP still has the time consuming problem. Island Model informed Genetic Algorithm was used in this research to solve this problem. The idea of this research is making distributed model exchanges an island's local best Individu with another island. Island model GA could create university course timetabling in reasonable time. This distributed model could run faster rather than single machine model dec reasing constraint violations to reach optimum fitness. It could have less constraint violations because it could escape from stagnant local optimum easier. Isl and model GA could even produce great accuracy for Telkom University dataset (99.74%) and acceptable accuracy at 96.80% for Purdue dataset for student level timetabling.