Akademska digitalna zbirka SLovenije - logo
E-viri
Recenzirano Odprti dostop
  • A Novel approach for predic...
    Zubaidi, Salah L.; Dooley, Jayne; Alkhaddar, Rafid M.; Abdellatif, Mawada; Al-Bugharbee, Hussein; Ortega-Martorell, Sandra

    Journal of hydrology, June 2018, 2018-06-00, Letnik: 561
    Journal Article

    •A novel model was developed to examine the impact of climate change on water demand for mid-term.•The model has been developed from Combining Singular Spectrum Analysis with Neural Networks.•Monthly climatic factors and water consumption were used as inputs and output in the model for the period (2006–2015).•The results present that the model is skilfully and reliable to predict water demand.•The findings of this study support the view that water demand is driven by climatological variables. Valid and dependable water demand prediction is a major element of the effective and sustainable expansion of municipal water infrastructures. This study provides a novel approach to quantifying water demand through the assessment of climatic factors, using a combination of a pretreatment signal technique, a hybrid particle swarm optimisation algorithm and an artificial neural network (PSO-ANN). The Singular Spectrum Analysis (SSA) technique was adopted to decompose and reconstruct water consumption in relation to six weather variables, to create a seasonal and stochastic time series. The results revealed that SSA is a powerful technique, capable of decomposing the original time series into many independent components including trend, oscillatory behaviours and noise. In addition, the PSO-ANN algorithm was shown to be a reliable prediction model, outperforming the hybrid Backtracking Search Algorithm BSA-ANN in terms of fitness function (RMSE). The findings of this study also support the view that water demand is driven by climatological variables.