基于不确定场定义模糊贡献度的多源测绘信息数据挖掘研究  

Multisource surveying and mapping information data mining based on fuzzy contribution degree of uncertainty field definition

作　　者：梁科 LIANG Ke(Loudi Geographic Information Center,Loudi 417000,China;Central South University,Changsha 410000,China)

机构地区：[1]娄底市地理信息中心,湖南娄底417000 [2]中南大学,湖南长沙410000

出　　处：《现代电子技术》2023年第18期71-76,共6页Modern Electronics Technique

基　　金：湖南省科技项目(202000560001)。

摘　　要：地理测绘信息来源具有多源性,不同数据源下的数据需要大量模糊语言描述,但缺少精准定义不同来源测绘信息的特征属性。针对不同来源的数据关系挖掘精准度差的问题,文中设计一种基于不确定场定义模糊贡献度的多源测绘信息数据库相似关系挖掘算法。使用K-means算法划分测绘多源数据类型,把划分结果作为训练数据集并转变为矩阵模式,代入反向传播(Back Propagation,BP)神经网络中训练,并清洗多源信息数据。利用数学中的不确定场模型定义测绘多源信息的模糊贡献度,获得数据库多源信息模糊水平。在遗传算法基础上引入兴趣度概念,定义两个测绘信息同时存在的信息量,调节染色体适应度,将杂交概率与染色体相似度的耦合关联作为多源信息模糊相似关系输出值,完成预期数据挖掘工作。实验结果表明,所设计的挖掘算法模糊相似关系挖掘时间约为7.5 s,方差值在6以内,数据挖掘结果的精准度在85%以上,可为地理测绘数据的正确应用提供可靠借鉴。The sources of geographic surveying and mapping information are multi-source.The data under different data sources need a large number of fuzzy language descriptions,and there is a lack of accurate definition of the characteristics of surveying and mapping information from different sources.In allusion to the poor accuracy in mining data relationships from different sources,a mining algorithm of similarity relationship in multi-source mapping information database based on fuzzy contribution degree defined by uncertainty field is designed.The K-means algorithm is used to divide the types of multi-source surveying and mapping data,and the division results are taken as the training data set and transformed into the matrix mode,which is then substituted into the BP(back propagation)neural network for training,and the multi-source information data is cleaned.The uncertainty field model in mathematics is used to define the fuzzy contribution degree of multi-source surveying and mapping information to obtain the fuzzy level of multi-source information in database.On the basis of genetic algorithm,the concept of interest degree is introduced to define the amount of information that two mapping information exist at the same time,the chromosome fitness is adjusted,and the coupling association between the hybridization probability and the chromosome similarity is used as the output value of the fuzzy similarity relationship of multi-source information to complete the expected data mining.The experimental results show that the mining time of the designed mining algorithm for fuzzy similarity relations is about 7.5 s,the variance is within 6,and the accuracy of the data mining results is above 85%,which can provide a reliable reference for the correct application of geographic surveying and mapping data.

关 键 词：地理测绘多源信息 模糊贡献度 模糊相似关系 数据挖掘 不确定场 遗传算法 

分 类 号：TN919-34[电子电信—通信与信息系统] TP391[电子电信—信息与通信工程]