A First Course in FUZZY and NEURAL CONTROL

212 CHAPTER 6. NEURAL CONTROL

net1 = init(net1);

net1.iw{1,1}

net1.b{1}

net2 = init(net2);

net2.iw{1,1}

net2.b{1}

The training period is set up for 10,000 epochs along with a mean-squared error
goal of 0.0001.

net1.trainParam.epochs = 10000;

net1.trainParam.goal = 0.0001;

net2.trainParam.epochs = 10000;

net2.trainParam.goal = 0.00001;

The networks ìnet1îandìnet2î can now be trained to the proportional gain
and integral gain functions developed previously.

net1 = train(net1,dp,tprop);

net2 = train(net2,p,tint);

The two trained neural networks ìnet1îandìnet2î are resimulated to check
how well the networks have learned the control functions, and the plots shown
in Figures 6.12ó6.15 are obtained.

-0 .4^0 -0 .2 0 0.2 0.4 0.6

1

2

3

4

5

6

Figure 6.12. Proportional gain versus gradient for ìnet1î