Evaluation of Psonns Model in Daily Flow Discharge Prediction (Case Study: Babolrood River, Iran)

Abstract

In recent years, artificial neural networks (ANNs) have been successfully used as a tool to model various non-linear relations, and the method is appropriate for modelling the complex nature of hydrological systems. Owing to the complexity of the hydrological process, Particle Swarm Optimization Neural Networks (PSONNs) is the superior model that is able to calibrate the daily flow discharge accurately by using only flow data. Therefore, a new evolutionary algorithm (EA) named particle swarm optimization (PSO) is proposed to train the feedforward neural networks. This particle swarm optimization feedforward Neural Networks (PSONNs) is applied to model the daily flow discharge for Babolrood river in Mazandaran a province in Iran. With the input data of antecedent flow discharge, the optimal configuration of PSONNs is able to simulate current flow discharge successfully with an accuracy of R2=0.683, MSE=0.0023 and MAE=0.0206 for training and R2=0.736, MSE=0.0024 and MAE=0.0206 for testing data set. The performance of the newly developed PSONNs demonstrated the success in modeling flow discharge for the Babolrood River

Keywords: ANNs, PSO, PSONNs, Babolrood river, Daily flow prediction