where : N is the number of the samples,I is the number of inputs,O is the number of outputs, Ntrn is the number of training equations , Neq is the number of equations. and Hub is a limit we put for limiting the number of retraining of the network. To have a Robust Network there are conditions that :



begin{equation}

begin{cases}
Hmax << Hub  \
Nw << Neq
end{cases}
end{equation}


Hidden layer numbers are altered from 1 to 15 , and the H = 13 is found to have the best results.

subsection{The results of the training }

The training function divides the data to three blocks : training, validation, and test.