The number of nodes in the hidden layer is determined
empirically as indicated above in the first method. It was
found that, the 6 hidden nodes gave the most precise
estimation and the result of this case as shown in Figure 4.
This illustrates the neural network construction in
MATALB/SIMULINK environment, the input layer has 2
nodes, the hidden layer has 6 nodes and the output layer has 1
node. Accordingly, the input layer consists of a two
dimensional vector, PV generator output power and the DC
motor speed. The output layer of the proposed ANN
comprises only, a one dimensional vector which is the
controlling signal of the buck chopper. The outputs of the
input layer are weighted and summed together, and then a
bias is added to the sum of the weighted inputs. Finally the
summation of the weighted inputs and the bias are passed
through the transfer function f. There are different types of
transfer function that are used as activation functions, such as
sigmoid, Gaussian, tangent, hyperbolic constant etc. Two of
the most commonly used functions, tangent and linear
function are used in this work.