| Summary: | 碩士 === 逢甲大學 === 統計與精算所 === 92 === This paper studies a two-machine flowshop scheduling problem with a learning effect where its objective is to find a sequence to minimize the maximum tardiness. We apply a branch-and-bound method and a simulated annealing (SA) method to search for an optimal solution and a near-optimal solution, respectively. The computational results show that the numbers of nodes explored increase sharply as the learning effect becomes stronger, and the data with shorter ranges of due dates are more difficulty than those with longer ones, and the mean of the ratio of optimal found from SA is greater than 96 percent and its worst case is still greater than 93 percent. Besides that, the comparison between the SA and the earliest due date first (EDD) rule on large-job sizes are also provided. The results indicate the improvement spread increases as the learning effect becomes weak.
|
|---|
