A short-term water demand forecasting model using multivariate long short-term memory with meteorological data

Abstract Sustainable management of water resources is a key challenge nowadays and in the future. Water distribution systems have to ensure fresh for all users an increasing demand scenario related long-term effects due climate change. In this context, reliable short-term forecasting model crucial optimal resources. This study proposes novel deep learning based on long memory (LSTM) neural networks forecast hourly demand. Due limitations using multiple input sequences with different time lengths LSTM, proposed developed two modules that process temporal data: first module aimed at dealing meteorological information second representing longer-term The dual-module structure allows multivariate selection inputs length. performance compared conventional multi-layer perceptron (MLP) seasonal integrated moving average (SARIMA) real case study. results highlight potential approach prediction, outperforming more approaches.