摘要
河湖划界是河湖长的巡查工作范围依据和河湖管理基础。针对现有河湖巡查定界方法的不足,立足江西省基层河湖长日常巡查工作现状,充分利用现有的河湖巡查历史数据,提出了一种基于改进密度峰值聚类的河湖巡查定界算法。该算法在马氏距离下基于自然最近邻方法计算相对密度,采用自适应聚类中心的选取和合并策略并利用凸包算法形成虚拟电子围栏,依此确立河湖巡查的有效边界。与几种经典聚类算法进行对比实验发现该算法具有优异的聚类性能。最后,以抚河流域河湖巡查真实场景为例,给出了该算法在历史巡河数据上的实际应用效果,进一步验证了该算法的可行性和有效性。
The demarcation of rivers and lakes serves as the basis for the inspection scope of river and lake chiefs and is fundamental to river and lake management.In order to address the shortcomings of existing methods for river and lake patrol boundary demarcation,the paper proposes an improved density peak clustering algorithm based on the current practices of grassroots river and lake chiefs'patrol work in the province,utilizing historical patrol data.This algorithm calculates relative density based on natural nearest neighbor method under Mahalanobis distance,adopts adaptive clustering center selection and merging strategy,and uses convex hull algorithm to form virtual electronic fence,thereby establishing effective boundaries for river and lake patrol.Comparative experiments with several classic clustering algorithms have shown that this algorithm has excellent clustering performance.Finally,taking the real scene of river and lake patrols in the Fuhe River Basin as an example,the practical application effect of the algorithm on historical river patrol trajectory data is presented,
作者
谢敏
周毅超
曾斌
罗莹
王鹏
王文丰
XIE Min;ZHOU Yichao;ZENG Bin;LUO Ying;WANG Peng;WANG Wenfeng(Jiangxi Flood Control Information Center,330009,Nanchang,PRC;School of Information Engineering,Nanchang Institute of Technology,330099,Nanchang,PRC)
出处
《江西科学》
2025年第1期10-17,共8页
Jiangxi Science
基金
国家自然科学基金项目(61962036)
江西省水利厅科技重点研发项目(202325ZDKT17,202426ZDKT13)。
关键词
巡河轨迹
马氏距离
自然最近邻
密度峰值聚类
越界巡查
river patrol trajectory
Mahalanobis distance
natural nearest neighbor
density peak clustering
cross-boundary patrol
