Fig. (2) Architecture of feed forward back propagation neural network. Network architecture of 6-25-1 representing the input, hidden and output layer was found to be optimum for the prediction of desired response.