面向积水推测的机会式感知轨迹选择  

Selection of Opportunistic Sensing Trajectories for Ponding Water Inference

作　　者：张伟杰 於志勇[1] 黄昉菀[1] 朱伟平[1] ZHANG Weijie;YU Zhiyong;HUANG Fangwan;ZHU Weiping(College of Mathematics and Computer Science, Fuzhou University, Fuzhou 350108, China)

机构地区：[1]福州大学数学与计算机科学学院,福建福州350108

出　　处：《郑州大学学报（理学版）》2021年第4期102-108,共7页Journal of Zhengzhou University:Natural Science Edition

基　　金：国家自然科学基金项目(61772136)。

摘　　要：城市积水严重影响了城市居民的日常出行和灾害天气下城市的正常运作。及时发现城市各处是否发生积水显得尤为重要,但是以往监测城市积水的方式多是通过人为反馈、设备监测等方式来实现的,这种方式覆盖范围小、成本较高且容易出错。对深圳市部分区域进行网格划分,融合深圳市滑动雨量数据、深圳市公交线路轨迹数据、深圳市水务局积涝点水位数据,并提取相关特征,使用孤立森林算法、压缩感知算法对所有积水监测站点的积水状态进行推测,最后结合群智感知,选取公交车来参与感知任务,采集积水数据提高推测准确度。The existence of urban ponding water in the city could greatly affect the daily travel of urban residents and the normal operation of the city under severe weather.Therefore,it was particularly important to find out whether there was urban ponding water in various parts of the city in time.However,in the past,the methods of monitoring urban ponding water were mostly achieved through human feedback,equipment monitoring and other methods with small coverage,high cost and error-prone methods.Some areas of Shenzhen were divided into grids,Shenzhen sliding rainfall data,Shenzhen bus line trajectory data,Shenzhen Water Affairs Bureau waterlogging water level data,and relevant features were extracted.Isolation Forest and Compressed sensing algorithm were used to analyze all the accumulations,the state of urban ponding water at the water monitoring site was estimated,and finally combined with Crowd-sensing,buses were selected to participate in the perception task to collect ponding data to improve the accuracy of the estimation.

关 键 词：城市积水 孤立森林 城市计算 群智感知 数据融合 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]