Hybrid Models for Water Demand Forecasting

Abstract

An accurate prediction of future water consumption is necessary to create a satisfactory design for a water distribution system. In this study, two new hybrid approaches are proposed for accurately predicting future hourly and monthly water demands. The first approach is based on the hybridization of ensemble empirical mode decomposition (EEMD) and difference pattern sequence forecasting (DPSF), and the second is based on the hybridization of EEMD with DPSF and autoregressive integrated moving average (ARIMA). Historical hourly water consumption datasets of southeastern Spain and monthly datasets of Nagpur, India are used for assessing the performance of the proposed approaches. The performance of the EEMD-DPSF approach is checked using the root mean square error (RMSE), mean absolute error (MAE), and mean percentage absolute error (MAPE). Further, the results are compared with those obtained using PSF, ARIMA, DPSF, their hybrid models, and various other ANN models. The proposed EEMD-DPSF method is found to perform significantly better than the other state-of-the-art methods in terms of prediction accuracy without compromising time and memory complexities. The comparison between the two proposed models demonstrates that the EEMD-DPSF approach provides better results, whereas the EEMD-DPSF-ARIMA approach requires shorter computational time.