Establishing an Optimization Model for Sewer System Layout with Applied Genetic Algorithm
Establishing an Optimization Model for Sewer System Layout with Applied Genetic Algorithm
复制标题
应用遗传算法建立下水道系统布局优化模型
DOI：
10.3808/jei.200500043
复制
发表时间：
2005
影响因子：
7
通讯作者：
S. Liaw
中科院分区：
文献类型：
--
作者：
H. Weng;S. Liaw
本研究首先利用遗传演算法建立一个组合最佳化模式，称为排水系统布局与水力学最佳化模式（GA/SSOM/LH），以寻求一个真实的城市排水系统的最佳化设计。同时考虑了“管网布置”和“水力设计”优化问题。建模的概念是联合收割机的基本原则的遗传算法，以产生可能的网络布局，以及开发一个“水力设计”优化模块，污水系统优化模型（SSOM），它可以找到最佳的污水系统布局，通过检查整体最低成本的水力设计的几个可能的替代网络布局。SSOM是一种0-1混合随机规划（MIP），其中应用了传统的有界隐式枚举（BIE）算法来确定每个的最佳大小和斜率。与BIE算法不同，GA进化中的“一个染色体”被编码为代表“一个系统布局”参数。“参数”上的特定编码字符串直接操作，与SSOM模块结合使用时更加稳健。因此，遗传算法可以迅速发展，产生一个优化的系统布局，并确保解决方案更接近全局最优的“快速”的方式。最后，一个73节点的项目进行了案例研究，以验证GA/SSOM/LH模型产生的最优系统布局。
In this study, a genetic algorithm (GA) is first used to establish a combinatorial optimization model, called the Sewer System Optimization Model for Layout & Hydraulics (GA/SSOM/LH), to find an optimal design for a real urban sewer system. The problems of "network layout" and "hydraulic design" optimization are considered simultaneously. The modeling concept is to combine the fundamental principles of the GA, to the generation of possible network layouts as well as to develop a "hydraulic design" optimization module, the Sewerage System Optimization model (SSOM), which can find the best sewer system layout by checking the overall least-cost hydraulic design of several possible alternate network layouts. SSOM is a 0-1 Mixed Integer Programming (MIP) in which a traditional algorithm, the Bounded Implicit Enumeration (BIE) is applied to determine the optimal size and slope for each. Unlike the BIE algorithm, 'one chromosome' in the GA evolution is coded to represent 'one system layout' parameter. Specific coding strings on 'parameters' are operated directly and are more robust when combined with the SSOM module. Hence the GA can evolve quickly generating an optimized system layout and ensuring a solution closer to the global optimum in a 'fast' manner. Finally, a case study was conducted on a 73-node project to verify the optimal system layout as generated by the GA/SSOM/LH model.