Document Type : Original Article
Authors
1
Department of Water Engineering, Kermanshah Branch, Islamic Azad University, Kermanshah, Iran.
2
Department of Water Engineering, Kermanshah Branch, Islamic Azad University, Kermanshah, Iran
3
Department of Water Engineering, Kermanshah Branch, Islamic Azad University , Kermanshah, Iran.
4
Department of Water Engineering , KermanshahBranch, Islamic Azad University,Kermanshah, Iran.
5
Department of Water E ngineering , Kermanshah Branch, Islamic Azad University, Kermanshah, Iran.
6
Department of Water Engineering, Kermanshah Branch , Islamic Azad University, Kermanshah, Iran.
Abstract
Objective: The aim of this study is to create a neural network model based on collective data encounter (GMDH) and improve it using Harmony Search Optimization (HSA) algorithm for simulating monthly river flow (HSA-GMDH).
Method: For this purpose, the rainfall and flow data of Gamasiab river of Kermanshah were used during a 20-year period (1370-1390). To develop the GMDH model, 80% of the data were used to train the model and other data were used to evaluate it. Also, the best input variables to the model were determined by trial and error method, based on this method, 3 datasets were formed (D1, D2, D3), then the performance of GMDH and HSA-GMDH models was evaluated with each of these datasets.
Results: The HSA-GMDH(D1) model performs better in the training and testing phase than the GMDH(D1) model. The HSA-GMDH(D2) model also performs better than the GMDH(d2) model. The HSA-GMDH(D3) model also performs better than the GMDH(D3) model. It is better to use an optimization algorithm such as HSA instead of the trial and error method to simulate the monthly flow of the river using the GMDH model.
Conclusions: Based on the results of this study, the HSA-GMDH model performs much better than the GMDH model, so it can be used as a powerful tool for simulating monthly river flow.
Keywords
Main Subjects
)