基于最优化的2008奥运会临时超市设计

摘 要

本文研究了2008年北京奥运会临时超市网点设计问题。这是1个最优化的实际应用问题。

根据3份问卷调查共10，000份数据，用EXCEL表统计出观众在出行，用餐和购物等方面规律，基于这些规律，结合题中已知信息，建立了商区人流量分布模型，消费期望值分布模型和网点设计模型。

在求解商区人流量分布模型和消费期望值分布模型时，引入了矩阵思想，通过编制MATLAB程序对其进行求解，简化了运算量，所得的结果表明：商区人流量分布比率与消费期望值有关，但不成比例关系；在求解网点设计模型时，考虑到这是1个多约束条件，多峰值的，大解空间的非线性规划问题，为避免1般优化算法的高维灾难问题，采用了基于搜索寻最优解思想的遗传算法，并设计了3组不同的MS值，所得的结果偏差较小，与实际较为贴近。其中，网点设计模型能适合更大规模问题的求解，通用性较强。

关键字：人流量分布模型；消费期望值分布模型；网点设计模型；矩阵；遗传算法。

Abstract

This article studies the designing problem of Beijing Olympic Games’ temporary supermarket network in 2008. It want to seek a optimization key actually.

According to three questionnaire research data whose number is over 10,000, we make use of the tool of EXCEL and have the rules of the audience in travel, having dinner and shopping ,Based on this rules and analyzed information known ,we have built people flowing distributing models in commercial district ,the distributing models of the expected consumption and the designing models of network.

In course of answering people flowing distributing models in commercial district and the distributing models of the expected consumption ,we make the matrix use of calculating ,and make the MATLAB program in answering the questions ,it reduces the amount of work ,The results show that: the rate of people flowing distribution in commercial district has relation to the expected consumption ,but they are disproportionate ;As calculating the designing models of the network, Because this is a excessive restrict conditions ,more peak values and big answer space nonlinearity layout problem ,In order to avoid high dimension tragedy issue that commonly optimize algorithm has, we adopt the heredity algorithm which based on searching furthest idealistic and contrive three different MS values, The results ‘s warp is less, and press close to fact. Specially, The designing models of the network can be fit to answer a more cosmically problem, and it has better currency.

Keywords： People flowing distributing models in commercial district ；The distributing models of the expected consumption; The designing models of network； Matrix；Hereditary algorithm。