基于BP神经网络的山地村分类方法研究——以重庆市巫溪县为例

A Classification Method of Mountainous Villages Based on BP Neural Network:A Case Study in Wuxi County,Chongqing City

《国家乡村振兴战略规划(2018-2022年)》明确提出,按照集聚提升、融入城镇、特色保护、搬迁撤并分类推进乡村发展。我国幅员辽阔,村数量庞大,传统的分类方法效率较低,且受人为因素干扰较大,随意性较强。本文以重庆市巫溪县村布局规划为例,从生态本底、土地承载、交通条件、自然灾害、经济人口、历史人文6个方面,选取13个学习指标(输入量)与2个筛选指标构成巫溪县村分类的指标体系,将巫溪县289个行政村依照不同比例分为训练样本和测试样本,通过BP神经网络的机器学习方法进行村分类,并对各类型村发展优先级测度进行计算。结果表明,BP神经网络对村类型的划分有较高的准确率(分类准确率为98.6%),并对不同类型村的发展提出时序性的指导,有效提高了村分类的效率和规划的科学性。

The National Strategic Plan for Rural Revitalization(2018-2022) proposes to promote rural development under the classification:agglomeration for improvement,integration into cities and towns,characteristic protection,relocation and merging.China is a vast country with a large number of villages.However,the traditional classification method is less efficient and more arbitrary due to the interference of human factors.Taking the village layout planning of Wuxi County,Chongqing City as an example,we selected 13 learning indicators(input quantity)and 2 screening indicators from 6 aspects:ecological background,land bearing,traffic conditions,natural disasters,economy,history and humanities to form an indicator system in Wuxi County.The village classification is carried out by BP neural network,and the development priority measure of each type of village is calculated.The results show that the BP neural network has a high accuracy rate(98.6%) in classifying village types,and provides development timing guidance for different types of villages.The method could effectively improve the efficiency of village classification and the scientificity of planning.