Monitoring And Prediction Of Urban Pipe Network Drainage State Based On SWMM And Neural Network

Due to extreme climate change and the acceleration of urbanization,urban waterlogging has become increasingly serious.The urban drainage pipe network is the main way of urban flood discharge.The drainage state of the pipe network is closely related to the occurrence and development of urban waterlogging.Monitoring and forecasting the drainage state of the drainage pipe network,such as node overflow,can effectively improve the disaster prevention and mitigation capabilities of urban waterlogging.In the traditional monitoring and forecasting process of drainage pipe network based on the Urban Rain and Flood Model(SWMM),the deviation of the forecasted rainfall is discrete,resulting in a large error in the forecast results.Aiming at this problem,this paper integrates SWMM model and BP neural network technology to establish an urban drainage pipe network status monitoring and forecasting model that does not require discrete SWMM coupled BP neural network for forecasted rainfall,and improves the accuracy of urban drainage pipe network node overflow monitoring and forecasting.The main research is as follows:According to the underlying surface factor,the generalized data of the drainage pipe network and the results of the division of the catchment area,the SWMM-based urban drainage pipe network simulation model is established.Taking the monitored pipe network state parameters as training samples,an urban drainage pipe network state prediction model based on BP neural network is established.The two models are coupled and connected to establish the SWMM coupled BP neural network urban drainage network state monitoring and prediction model,and the SWMM coupled BP neural network model and the prediction accuracy based solely on the SWMM model are compared and analyzed.The results show that:under the condition of small precipitation,the prediction result of the model based solely on SWMM is significantly smaller,and the accuracy of both models will increase with the increase of precipitation.Compared with the prediction results based on the SWMM model,the SWMM coupled BP neural network model reduces the root mean square error(RMSE)by 20.23m~3,the average relative error(MRE)by 18.06%,and the average absolute error(MAE)by 16.08m~3.The SWMM-coupled BP neural network-based urban drainage pipe network state monitoring and prediction model is superior to the SWMM-based urban drainage pipe network state simulation model,which improves the accuracy of node overflow monitoring and prediction.