Application of Artificial Neural Network (ANN) for the Modelling and Prediction of Ikpoba River Flow Data

Abstract

This study focuses on evaluating the effectiveness of artificial neural network (ANN), specifically employing the Levenberg Marquardt Back Propagation algorithm, for streamflow prediction in Ikpoba River, Benin City. Hydrological data from September 2022 to March 2023, along with historical data (2010-2015), were used to establish a cubic polynomial relationship between gage height and river discharge. The ANN model, with 10 hidden neurons, demonstrated a strong performance with a mean square error of 0.000303, surpassing the target error of 0.01. The study highlights the significance of accurate streamflow prediction models, especially in mitigating floods and optimizing reservoir management.