耦合水文和神经网络的城市积水动态风险评价  

Dynamic risk assessment of urban stagnant water with coupling hydrological and neural network

作　　者：刘野 逯跃锋[1,2] 孟庆祥 秦永[1] 刘聪[1] 闻俏 胡玉龙 LIU Ye;LU Yuefeng;MENG Qingxiang;QIN Yong;LIU Cong;WEN Qiao;HU Yulong(Shandong University of Technology,Zibo,Shandong 255049,China;Institute of Geographical Sciences and Natural Resources Research,Chinese Academy of Sciences,Beijing 100101,China;Wuhan University,Wuhan 430079,China;China Transport Telecommunications&Information Center,Beijing 100011,China)

机构地区：[1]山东理工大学,山东淄博255049 [2]中国科学院地理科学与资源研究所,北京100101 [3]武汉大学,武汉430079 [4]中国交通通信信息中心,北京100011

出　　处：《测绘科学》2020年第8期174-180,共7页Science of Surveying and Mapping

基　　金：国家自然科学基金项目(91646207);国家重点研发计划项目(2017YFC0822000,2017YFC1405000)。

摘　　要：为了综合分析城市积水点的时空动态及其风险等级,该文将水文模型与深度神经网络(DNN)相结合,提出了耦合水文模型和神经网络的城市暴雨积水点的风险评价模型。应用水文模型计算降雨过程中积水点的动态积水量以及积水区域;使用缓冲区分析来提取积水点附近道路的兴趣点(POI)。然后利用积水量、积水范围和POI的信息,采用深层神经网络对积水点风险进行评价;利用贵阳市中心城区的内涝调查数据对模拟结果进行实验验证。结果表明,该模型能够比较精确地评估城市积水灾害风险等级及其时空分布。In order to comprehensively analyze the spatiotemporal dynamics and risk level of urban water accumulation points,the hydrological model was combined with deep neural network(DNN),and a risk assessment model of urban stormwater accumulation points coupled with hydrological model and neural network was proposed.The hydrological model was used to calculate the dynamic water accumulation and the water accumulation area of the water accumulation point during the rainfall process.And buffer analysis method was used to extract points of interest(POI)for roads near the water accumulation points.Then,the deep network was used to evaluate the water accumulation point risk by using the information of water accumulation,water accumulation range and POI.Finally,this paper used the intrinsic survey data of the downtown area of Guiyang to verify the experiment.The results showed that the model could accurately assess the risk of urban water disasters and their spatial and temporal distribution.

关 键 词：积水点 风险评估 水文模型 深度神经网络 

分 类 号：P208[天文地球—地图制图学与地理信息工程]