運用網路方法於晶圓針測排程規劃之研究

Abstract

As the increasing requirements of electronic products, the electronic components are strongly needed for product composing. Accordingly, many enterprises devote themselves to manufacture all necessary electronic elements and attempt to pursue large fortune. However, it is hard to emerge first from the competitive industrial. Handle the key competition well is the main idea to be the champion of many excellent industries. Besides the basic consideration of controlling product quality, product due date is another critical criterion for the customers choosing product providers. Consequently, the excellent yield controlling coordinating the highly accurate due date can be the key-point for retaining customers. Comparing with the factors affecting the due date accuracy, the production lead-time plays a more essential role. Good production planning and scheduling maintain the production lead-time in control. However, before constructing a better production planning, the characteristics of wafer probing must be considered to improve the due date accuracy. The characteristics contain dynamic lots arrival, sequentially dependent setup time, production reentrant processing. Besides, due to the high price of test equipments, cost saving is another crucial task to avoid capacity abuse. As a result, how to fully utilize machine capacity by minimizing machine setup time is the crucial in this thesis. Owing to many scheduling characteristics and the satisfaction of shorten production lead time, a module of production planning and scheduling is constructed. Moreover, the network algorithms are used to schedule the due date oriented job processes based on fully utilization of machine capacity.At first, an integer programming model is constructed to optimize the production schedule in the consideration with the wafer probing characteristics, which must be analyzed and generalize first. However, the IP model is complex to describe real-world job processing condition and there would be a long time to optimize scheduling solutions. As the result, network algorithms are adopted in this thesis. Based on the network algorithms, several solving index is used to plan the reasonable and practicable scheduling. Considering several processing characteristics of wafer probing, the network algorithms are modified to apply on the real-world scheduling problem.

As the increasing requirements of electronic products, the electronic components are strongly needed for product composing. Accordingly, many enterprises devote themselves to manufacture all necessary electronic elements and attempt to pursue large fortune. However, it is hard to emerge first from the competitive industrial. Handle the key competition well is the main idea to be the champion of many excellent industries. Besides the basic consideration of controlling product quality, product due date is another critical criterion for the customers choosing product providers. Consequently, the excellent yield controlling coordinating the highly accurate due date can be the key-point for retaining customers. Comparing with the factors affecting the due date accuracy, the production lead-time plays a more essential role. Good production planning and scheduling maintain the production lead-time in control. However, before constructing a better production planning, the characteristics of wafer probing must be considered to improve the due date accuracy. The characteristics contain dynamic lots arrival, sequentially dependent setup time, production reentrant processing. Besides, due to the high price of test equipments, cost saving is another crucial task to avoid capacity abuse. As a result, how to fully utilize machine capacity by minimizing machine setup time is the crucial in this thesis. Owing to many scheduling characteristics and the satisfaction of shorten production lead time, a module of production planning and scheduling is constructed. Moreover, the network algorithms are used to schedule the due date oriented job processes based on fully utilization of machine capacity.At first, an integer programming model is constructed to optimize the production schedule in the consideration with the wafer probing characteristics, which must be analyzed and generalize first. However, the IP model is complex to describe real-world job processing condition and there would be a long time to optimize scheduling solutions. As the result, network algorithms are adopted in this thesis. Based on the network algorithms, several solving index is used to plan the reasonable and practicable scheduling. Considering several processing characteristics of wafer probing, the network algorithms are modified to apply on the real-world scheduling problem.