Prediction of multi-components (chlorine, biomass and substrate concentrations) in water distribution systems using artificial neural network (ANN) models

Abstract

Artificial Neural Networks (ANN) models are used to predict residual chlorine, substrate and biomass concentrations in a Water Distribution System (WDS). ANN models with different architectures are developed: a one output ANN model (predicting chlorine, substrate and biomass individually), a two output ANN model (predicting chlorine + substrate, chlorine + biomass or substrate + biomass) and a three output ANN model (chlorine + substrate + biomass). This study is carried out for the Bangalore City and North Marin WDSs. Data for these WDSs is obtained from the multi-component reaction transport model. The models are compared using the correlation coefficient (R) and the Mean Absolute Error (MAE). The models developed are able to predict, reasonably well, the temporal variations in the chlorine, substrate and biomass concentrations. Error analysis is carried out to determine the robustness of the models.