A First Course in FUZZY and NEURAL CONTROL

206 CHAPTER 6. NEURAL CONTROL

p(i)=(exp(-time))*sin(w*time);

time=time+0.1;

if (p(i)>0.0)

t(i)=1;

elseif (p(i)<0.0)

t(i)=-1;

end

end

Figure 6.3 illustrates the damped sinusoid generated from the precedingMat-
labcode.

Figure 6.3. Damped sinusoidal pattern for neural network training

Figure 6.4 illustrates the relay characteristic obtained by plotting the pattern
vectorp(i)versus the targett(i). The neural network will be trained to repro-
duce this relay characteristic.

-1-0 .6 -0 .4 -0 .2 0 0.2 0.4 0.6 0.8

-0 .8

-0 .6

-0 .4

-0 .2

0

0.2

0.4

0.6

0.8

1

Figure 6.4. Plot of pattern vector versus target that represents a relay

characteristic

In our example, a two-layer feedforward network is created. The topology
of the network will comprise a hidden layer with some specified number of
neurons, and an output layer with as many neurons as we desire to represent