基因演算法在最適用水及處理網路設計上的應用

• Main Author: 蔡明政
• Other Authors: 張玨庭
• Format: Others
• Language: zh-TW
• Published: 2000
• Online Access: http://ndltd.ncl.edu.tw/handle/08230123523813949039

碩士 === 國立成功大學 === 化學工程學系 === 88 === This research solves the design problem for water usage and treatment networks (WUTN) with genetic algorithms. In the past, the task of water allocation was accomplished usually on the basis of experience. This author tries to integrate all water-using units, wastewater treatment units, water sources and sinks in a chemical process with one single mathematical programming model. It was found that an optimal WUTN could be synthesized easily with such an approach. Since then convergence rate of search process is often slow with the conventional genetic algorithms, several new features have been introduced. In particular, a special encoding technique has been developed to overcome the present optimization problem. The performance of GA computation is enhanced by introducing a so-called “inducing” parameter in a cascade evolution procedure. On the other hand, the novel concept of mixer and repeat treatment units has also been utilized to add an extra degree of freedom in WUTN design. As a result, the objective value can be further reduced without using none efficient units. Finally, the proposed techniques have also been applied to optimize a refinery located in southern Taiwan. Significant improvements were identified in this practical study.