利用离散霍普菲尔德神经网络解决基于"繁华度"的网点设计

针对2008年奥运会实际情况进行预测,首先从三次的调查结果中寻找规律,运用动态规划,逐步搜索最终得到最短路径,并求出各个商区的人流量。综合了人流量及该人流量的平均购买力来定义一个新的量:"繁华度",从而求出分配给各个商区的销售额。利用对商圈的研究的结论,得到用最少的商店满足最大人流需求的方案,根据克里斯塔勒的"中心地理论"和各个商区的人均消费水平,运用神经网络迭代并行处理优化;进一步求解,得到经过修正的大小规模商店个数,从而使结果更能符合实际预测2008年的情况。设计计算机仿真模拟算法,证明采用繁华度比例来分配商区规模是合理的。

Solution of Based-on-"prosperity" Net Point Design by Discrete Hopfield Neural Networks

Focusing at predicting the reality in 2008 Olympic Games, the regularity is finded from the 3 researches, searches step by step by dynamic programming, gets the shortest route and people current at each business district. A new variable is thus defined: "prosperity", which is based on the people current and the purchasing power per head. The sales amount in each business district depends on the prosperity. Using the conclusion of the research on business circle, a scheme of satisfying the maximum people current need with least stores is got. Considering the "Central Place Theory" by Christaller and the consumption per head, optimizing by iterative parallel processing by neural networks and solving further, the modified number and scale of stores are got. In this way the result is powerful in predicting the reality in 2008 O-lympic Games. Design a computer analog simulation algorithm to prove that it is reasonable to distribute business district according to prosperity.