Prediction And Prevention Of Urban Drainage Overflow Based On Neural Network

After the release of the notice of the State Council on printing and distributing the action plan for water pollution prevention and control in 2015,the upsurge of urban water environment treatment has been set off all over the country.Urban sewage collection system is an important infrastructure to ensure urban public health safety and urban water ecological health.With the rapid development of domestic urbanization,many problems of urban sewage collection facilities construction lag gradually appear.In the areas with poor management of combined system and separate system,the water level of sewage pipe network in southern cities is on the high side and even overflowing.How to accurately predict the water level of inspection wells in urban drainage network,so as to make the urban drainage management department have the ability to efficiently and quickly deal with the potential overflow problem of drainage network,and minimize the possible social impact caused by overflow,is the purpose of this research.Taking Guangzhou,a typical southeast coastal city,as an example,this study selects Xiaoguwei Island,where Guangzhou University Town is located,as the research area,analyzes and studies the natural geography,social economy,sewage system,rainwater system and various factors of sewage system operation of Xiaoguwei Island,and based on the existing urban drainage information system,analyzes and optimizes the layout of key nodes of sewage trunk pipe in the island After collecting the online monitoring data of the sewage system in 2020,the BP neural network model is constructed to simulate and predict the water level of the inspection well of the sewage pipe network in the island.Finally,the main reasons for the high water level and overflow of the pipe network are analyzed combined with the urban construction,and the corresponding preventive measures are put forward.The main conclusions are as followsIn Xiaoguwei island area,under the influence of tide level and flood,the receiving river of rainwater will support the rainwater system in the island and even back flow to a certain extent;in the combined drainage system and the diversion system with mixed connection of rainwater and sewage,rainfall will affect the water level of the sewage system;in the four urban villages of the island,there is incomplete diversion area,and a large amount of mixed sewage enters into the sewage system during heavy rainfall Water system will affect the water level of sewage system.Based on the urban drainage information system,nine well locations and four branch well locations of the main sewage collection pipe in the outer ring are selected to install online water level monitoring equipment.The water level data of 13 monitoring points and one rainfall measuring station in 2020 are collected continuously and uploaded to the urban drainage information system,which lays the foundation for the subsequent modeling and analysis.aking the annual data of 2020 as the analysis object,this paper qualitatively analyzes the factors affecting the water level of sewage pipeline,constructs BP neural network model,takes 70% of the annual data as the training sample and 30% as the test sample,determines 22 hidden neurons after repeated trial calculation,selects trainlm function as the network training function,learnlm function as the network learning function,and selects logsig Tansig group after comparative test Cooperation is the transfer function,and 0.001 is taken as the minimum value of network convergence according to the simulation results.BP neural network prediction results show that the prediction method can accurately predict the water level of sewage pipeline.The two main factors that affect the water level of sewage pipe are rainfall and tide level of outer river.Considering these two main factors,the paper analyzes that the main reasons for high water level and overflow of sewage pipe network are the mixture of rainwater and sewage and the invasion of external water,and puts forward some prevention and control measures,such as local reconstruction of municipal roads,reconstruction of balcony drainage in old residential areas,retention of rainwater system in urban villages,new sewage system and anti backflow of drainage outlets measures.To sum up,rainwater mixing,groundwater intrusion and river water body backflow are the main causes of high water level and overflow in urban sewage pipe network;the BP neural network model constructed in this study can accurately predict the water level of inspection wells in pipe network,and can provide reference for efficient management and emergency rescue of urban drainage management department,which has certain practical significance in drainage management.To sum up,rainwater mixing,groundwater intrusion and river water body backflow are the main causes of high water level and overflow in urban sewage pipe network;the BP neural network model constructed in this study can accurately predict the water level of inspection wells in pipe network,and can provide reference for efficient management and emergency rescue of urban drainage management department,which has certain practical significance in drainage management.