Assessment of neuro-fuzzy approach based different wavelet families for daily flow rates forecasting

Heavy rainfall over a short period or slowly during long periods can significantly increase the amount of water. Where it results in floods that can pose a direct threat, capable of causing human and material losses. Over the past two decades, artificial intelligence has been widely applied in the field of hydrology as well as in many other areas of hydraulic engineering. The wavelet transform is a very popular technique in the analysis of non-stationary time series and particularly effective for hydrological series. Currently, the application of intelligent hybrid systems in different fields has shown good performance and unparalleled efficiency. As such, in this work, we propose a hybrid neuro-fuzzy-wavelet model to modelling the rainfall-runoff transformation for forecasting daily flow rates in the Sebaou basin located in Tizi Ouzou region. The results obtained are very encouraging and better than those obtained by the models used for comparison in this research. According to the results, the hybrid neuro-fuzzy-wavelet model with mother wavelet db7 gave us better performance for daily flow rates forecasting.