基于深度学习的城市内涝预警系统设计

Design of Urban Waterlogging Warning System Based on Deep Learning

为降低城市内涝预警的误报率,解决由此产生的负面问题,本文提出基于深度学习的城市内涝预警系统设计。参照云计算架构,设计预警系统架构;为实现对降雨量的感知,完成硬件设备选型;引进深度学习技术,构建城市内涝数学模型;利用社会网络采集图像与城市降雨信息,选用Zigbee通信作为降雨数据主要通信方式;采用开放源GeoServer作为GIS中间件,设计雷达估算降水,完成城市内涝预警。实验结果证明:设计的基于深度学习的预警系统的误报率更低,即对于城市内涝灾害的预警精度更高,有一定的实际应用意义。

In order to reduce the false alarm rate of urban waterlogging early warning and solve the resulting negative problems,this paper proposes the design of urban waterlogging early warning system based on deep learning.Design the early warning system architecture with reference to the cloud computing architecture;To realize the perception of rainfall,complete the selection of hardware equipment;Introducing deep learning technology to build a mathematical model of urban waterlogging;Using social network to collect images and urban rainfall information,Zigbee communication is selected as the main communication mode of rainfall data;The open source GeoServer is used as GIS middleware,and radar is designed to estimate precipitation,thus completing urban waterlogging early warning.The experimental results show that the designed warning system based on deep learning has a lower false alarm rate,that is,it has a higher warning accuracy for urban waterlogging disasters,and has certain practical application significance.