Research on urban waterlogging risk prediction based on the coupling of the BP neural network and SWMM model

Abstract

Scientific and effective urban waterlogging risk prediction can help improve urban waterlogging disaster prevention capabilities. Combining the numerical simulation model with the data-driven model, the construction of the urban waterlogging risk predictive model can satisfy the prediction accuracy and improve the prediction timeliness. Thus, this paper established an urban waterlogging risk predictive model based on the coupling of the BP neural network and SWMM model, and set five input patterns, finally selected the accumulative precipitation process and precipitation characteristics as input to predict the regional waterlogging risks under different urban rainstorm scenarios. The results show that the overall performance of the pipe drainage system in the study area is lower, and it cannot resist the rainstorm with a higher return period. Moreover, the total waterlogging risk of the southern old city is higher than that of the northern new city in the study area. The calculation speed of the prediction model constructed in this paper is thousands of times higher than that of the numerical model, so the calculation speed is very fast, which meets the requirements of the forecast timeliness. HIGHLIGHTS Using fine data to build the SWMM model and various methods to verify the SWMM model.; Using web crawler technology to extract ponding points.; Using entropy weight method for risk quantization.; Using set pair analysis to evaluate the accuracy of the BP neural network.; Coupling SWMM model with BP neural network for risk prediction.;

Keywords

• bp neural network
• risk prediction
• swmm model
• urban waterlogging