Parallel Solving Feasibility Problems to Minimize Makespan for LED Sorting Operations

• Main Authors: Li, Chang-Feng, 李長峰
• Other Authors: Chen, Sheng-I
• Format: Others
• Language: en_US
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/5b7p8p

碩士 === 國立交通大學 === 工業工程與管理系所 === 106 === A LED wafer usually contains LED dies with distinct grades due to unstable quality in the manufacturing process. The main purpose of the sorting operation is to transfer LED dies to different bins according to their quality levels. A sorter plant usually involves of multiple machines with identical capability. The decision is to assign wafers to machines to minimize the makespan of sorting and merging times. To tackle this challenge, we construct a Mixed Integer Programming (MIP) model. A parallel computation algorithm is proposed by transforming the original problem into upper-bound feasibility problems and then evaluates each sub-problem in parallel. We further propose a hybrid method to enhance the efficiency when evaluating each feasibility problem. Explicit methods include reducing the problem size by substituting decision variables with promising solution, recovering the infeasible sub-problem by rolling back to its ancestor problem and then adding valid inequalities, and applying regularization for dealing with long-run time sub-problems. Computational results show that our proposed MIP model and solution approaches provides a rapid and robust decision for real-world problems.