Application of Genetic Algorithms to Optimization of Operator Scheduling Problem in Semiconductor Industries

• Main Authors: Kuan-Ju Lai, 賴冠汝
• Other Authors: 廖宜恩
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/vjnvs9

碩士 === 國立中興大學 === 資訊科學與工程學系所 === 101 === In semiconductor industry, mostly Fabs should have twenty-four hours production operation. There is a long-standing industry issues about how to both keep the normal operation of the factory and consider the vacation (or off days) flexibility of the operators. Therefore, the operators shift arrangements are very important. In the past, the shift schedules were artificially dispatched by the shift schedule supervisor. That is, the operators arrange vacation (or off days) what they want by themselves, and then the shift schedule supervisor gives the final decision according to the specified shift schedule policies and manpower. A good shift schedule arrangement not only must take into account the factory operations, but also must respect the autonomy of the employees. It is usually unable to strike a balance. If the shift schedule arrangement is short of careful considerations, it will affect vacation (or off days) of employees and the factory production, and result in a lose-lose situation. The manual shift scheduling costs in complexity and time for the shift schedule supervisor. Therefore, this study proposed an automatic shift scheduling system to replace the manual shift schedules. The proposed system considers both the operation of the factory and the vacation (or off days) flexibility of the operators, in order to obtain an optimized scheduling table. This study uses the core architecture of the genetic algorithm. With the characteristics of the semiconductor industry, and under the premise of the specified limitations (i.e., compliance with government regulations, policies of company, regulations of department, and personnel preferences etc.), the proposed system is evaluated by the following two objectives: 1. From the rights of employees: the proposed system should meet the work autonomy and fairness of vacation scheduling of employees. From a human resource management perspective, if employees think they have work autonomy and vacation scheduling fairness, it can improve the centripetal force and staff stability in a company. 2. From the benefits of company: the proposed system should meet the objective of maximizing the production and find a suitable shift schedule to reduce extra cost on overtime. Experimental results show that the proposed automatic shift scheduling system can reduce the shift scheduling time from 3-5 days by manual scheduling to 40 minutes. Moreover, by varying different weights on employee’s rights and company’s benefits, the proposed system gained about 45% to 62% improvement on the objective function compared to the manual scheduling method.