摘要
正确引导共享单车用户规范停车,是政府管理的重要举措。为此提出了利用聚类的原理、基于K均值的共享单车入栏检测算法。首先模拟产生服从高斯分布的随机单车定位数据;其次对定位数据进行聚类分组,不断迭代向均值移动确定每簇的中心点;最后将数据量最多的簇中心点作为单车的停放位置,对其进行入栏检测分析。用实际测得的单车定位数据验证入栏检测准确率,结果表明,在符合政府管理要求的条件下可准确检测单车是否入栏,正确率高达80%～100%,具有较好的实用价值。
In order to standardize parking,city managers mark parking areas with rectangular solid white lines on the pavement to remind riders to park their bicycles within the fence.A new method uses the K-means clustering principle for fence of shared bicycles is proposed.First,the location data of Gauss distribution are simulated.Second,to get multiple clusters,the data of each location are clustered,and then by means of continuous iteration,the final center of each cluster is determined.Finally,the data of the largest cluster of central points for entry fence detection are analyzed.The accurate rate of entry fence detection is verified by the actual measurement of location data of shared bicycles.The results show that the algorithm can accurately detect whether the shared bicycles enter the fence under the prescribed conditions,accuracy is as high as 80%-100%,and has high practical value.
作者
朱江淼
张菁
黄艳
金森林
高春柳
Zhu Jiangmiao;Zhang Jing;Huang Yan;Jin Senlin;Gao Chunliu(Information Department,Beijing University of Technology,Beijing 100124,China;Beijing Scientific Research Institute of Metrology and Testing,Beijing 100029,China)
出处
《电子测量技术》
2019年第8期98-103,共6页
Electronic Measurement Technology
基金
"空间导航与定位NQI技术集成及应用示范"项目(2017YFF0212000)资助
关键词
共享单车
围栏
GNSS高精度接收机
K均值聚类
入栏检测算法
shared bicycles
fence
high precision receiver of GNSS
K-means clustering
algorithm for entry fence detection
