Volume No. 4 Issue No.: 2 Page No.: 440-448 Oct.-Dec 2009

 

MULTI-SITE ESTIMATION OF SUSPENDED SEDIMENT LOAD BY ARTIFICIAL NEURAL NETWORKS

 

Nazmara H.*, Froutan Sh.1, Hosseinzadeh H.2 and Majidi A.3

1. National Iranian Oil Engineering and Construction Company (NIOEC), (IRAN)
2. Iran University of science and Technology, (IRAN)
3. Amirkabir University of Technology, (IRAN)

 

Received on : July 27, 2009

 

ABSTRACT

 

Sediment transport in a river can cause significant damages to the nature, agriculture and water installations. Rivers capacity reduction is the major reason of flooding which usually causes high damages to the environment and human manufactured. Therefore investigation about sediment transport and its modeling have dominant importance. In this paper, artificial neural networks are used in order to estimate suspended sediment load of Akhoola hydrometric station which is located by the side of the Ajichay River in east Azarbaijan province of Iran. The available data for the hydrometric stations are daily discharge and average sediment load. In the current study, feed forward back propagation network is used for sediment estimation. The effects of various factors such as logarithmic data as the networks input, normalization ranges, training algorithms and the number of hidden layers, on the model efficiency are inspected for obtaining the best results. In order to exanimate the effect of the upstream stations loads (i.e. Markid and Vanyar) on the Akhoola station load, the data of Markid and Vanyar stations are used for neural networks training in which presents the best result for sediment estimation of Akhoola station. Furthermore, sediment rating curves and linear regression models are also used in order to estimate the sediment load of the station. Genetic algorithm is used for optimization of the regression coefficients of the sediment rating curves, it is observed that genetic algorithm approach has no more advantages. In comparison with classical parameter estimation methods results, multi site estimation of the sediment by artificial neural networks can lead to the desired results.

 

Keywords : Artificial Neural Networks, Environment, Suspended Sediment load, Multi site Estimation, Ajichay River

 

 

BACK