Hybrid Multi-Offspring Genetic Algorithm for Scheduling Optimization of Semiconductor Manufacturing

Abstract

With the ongoing rise in demand for semiconductor-related technologies such as intelligence computing self-driving cars and automobile electronics Taiwan plays a pivotal role in the world ’s semiconductor industry Semiconductors have become the main force for Taiwanese economic growth Therefore in addition to state-of-the-art manufacturing technology how to optimize our production planning and scheduling to minimize makespan is also an important issue which is called job-shop scheduling problem In this study we proposed a new genetic algorithm Hybrid Multi-Offspring Genetic Algorithm (HMOGA) to improve the efficacy of iterative search by increasing the diversity of offspring and tested it by TSPLIB data Finally we used a job-shop scheduling problem on a two-stage hybrid flow shop with multiple processors (FSMP) of a 12-inch semiconductor fabrication plant in Southern Taiwan Science Park as an example to evaluate the performance of HMOGA by simulations According to the results HMOGA only requires 40-50% of the traditional genetic algorithm in computation time within 1 minutes and it can obtain a lower makespan that is superior to the traditional genetic algorithm and the dispatching rules Furthermore the appropriate mutation rate of HMOGA in this study is 0 1 ~ 0 3

Date of Award	2020
Original language	English
Supervisor	Mi-Chia Ma (Supervisor)