基于多源数据分析的区域划分形态识别与优化技术研究  

Research on region partition morphological recognition and optimization technology based on multi⁃source data analysis

作　　者：吴靖 WU Jing(Nanning Architectural and Planning Design Group Co.,Ltd.,Nanning 530002,China)

机构地区：[1]南宁市建筑规划设计集团有限公司,广西南宁530002

出　　处：《电子设计工程》2023年第14期57-61,共5页Electronic Design Engineering

摘　　要：传统的城市区域功能识别通常采用移动轨迹分析、POI兴趣点追踪技术来实现,手段单一且准确率较低。针对此问题,文中提出了一种基于多源数据融合分析的城市区域功能识别算法。该算法利用人群活动多源数据与自然语言文字处理的对应关系,使用Word2vec训练网络对人群活动数据进行训练。对于传统K-means聚类算法所存在的不足,从多个方面对其进行改进,改进后算法的聚类效果更优,同时还具有一定的隐私性,可以有效保证数据安全。实验测试结果表明,所提算法能够识别出的区域种类在对比算法中最多,且训练网络的准确率高达90.5%,能够实现城市区域功能的准确识别。The traditional urban area function recognition is usually realized by moving trajectory analysis and POI interest point tracking technology,which has a single means and low accuracy.In order to solve this problem,this paper proposes an algorithm of urban area function recognition based on multi⁃source data fusion analysis.The algorithm uses the corresponding relationship between multi⁃source data of crowd activity and natural language word processing,and uses Word2vec training network to train the crowd activity data.For the defects of the traditional K-means clustering algorithm,it is improved from many aspects.The improved algorithm has better clustering effect and privacy,which can effectively ensure data security.Experimental results show that this algorithm can identify the most types of regions in the comparison algorithm,and the accuracy rate of the training network is as high as 90.5%,which can realize the accurate identification of urban regional functions.

关 键 词：多源数据分析 区域识别 连续词袋模型 K-MEANS 数据聚类 城市规划 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TN99[自动化与计算机技术—计算机科学与技术]