Urban Drainage Network Overflow Risk Prediction Based on Fuzzy Brain Emotional Learning

The Fuzzy Brain Emotional Learning Classifier model (FBELC) was used to predict the overflow risk of drainage network in response to the increasingly serious urban waterlogging problem. Using 14 characteristic values such as catchment parameters, pipe network parameters and rainfall data as inputs, the FBELC parameters were updated by adaptive learning rules, and finally the risks were classified from I to IV. FBELC is compared with radial basis function (RBF) neural network, backpropagation (BP) neural network and fully connected neural network (FCNN). The method proposed in this paper can accurately predict the overflow risk level of drainage network, and the final accuracy rate is 92%. The FBELC model, based on total rainfall, achieved 96% accuracy in predicting the level of overflow risk for actual rainfall.