The artificial neural network (ANN) approach, which is a non-linear ...

glibdoadingΤεχνίτη Νοημοσύνη και Ρομποτική

20 Οκτ 2013 (πριν από 4 χρόνια και 2 μήνες)

108 εμφανίσεις



Artificial Neural Networks in Water Resources

H.Kerem Cigizoglu

Associate Prof., Hydraulics Division, Civil Engineering Faculty,
Istanbul Technical University,
I
stanbul, Turkey


Key words :

artificial neural network, hydrologic time series, forecasting,
feed forward back
propagation, radial basis functions, generalized regression neural network

Abstract

The artificial neural network (ANN) approach, which is a non
-
linear black box model,
would seem to be a useful alternative for modeling the
water resource
s time

series. The ANN
applications in water resources are in river flow prediction, in the rainfall
-
runoff relationship,
in rainfall estimation
,

in various groundwater problems

and in suspended sediment modelling
.

In the majority of these studies, the fe
ed
-
forward error back
-
propagation method (FFBP) was
employed to train the neural networks. The performance of the FFBP was found to be
superior to conventional statistical and stochastic methods in continuous flow series
forecasting. However, the FFBP algo
rithm has some drawbacks such as the local minima
problem.
In this study relatively new ANN methods such as
the radial basis function

and
generalized regression neural network are employed in the modeling of hydrologic time series
together with FFBP method
. The results are compared with those of the conventional
statistical methods.