Short-term water demand predictions coupling an artificial neural network model and a genetic algorithm

Abstract The application of artificial neural network (ANN) models for short-term (15 min) urban water demand predictions is evaluated. Optimization the ANN model's hyperparameters with a genetic algorithm (GA) and use growing window approach training model are also results compared to those commonly used time series models, namely Autoregressive Integrated Moving Average (ARIMA) pattern-based model. evaluations based on data sets from two Canadian cities, providing 15 min consumption records over respectively 5 years 23 months, respective mean 14,560 887 m3/d. GA optimized performed better than other Nash–Sutcliffe Efficiencies 0.91 0.83, relative root square errors 6 16% City 1 2, respectively. this study indicate that optimization an can lead predictions, which useful many real-time control applications, such as dynamic pressure control.