Information Technology Reference
In-Depth Information
a
100
90
80
70
60
50
40
30
20
5
10
15
20
time (h)
b
100
90
80
70
60
50
40
30
20
10
20
30
40
50
60
70
time (h)
Fig. 12.26
Approximation of the system's temperature output (
red line
) by a neural network with
Hermite polynomial basis functions (
blue-line
)(
a
) temperature's time variation—profile 1 (
b
)
temperature's time variation—profile 2
where e.k/ D y.k/ y
d
.k/ is the output estimation error at time instant k and
T
.k/ is the regressor vector having as elements the values .x.k// of the Gauss-
Hermite basis functions for input vector x.k/.
To approximate the temperature-signal's variations described in a data set
consisting of 870 quadruplets of input-output data, a feed-forward neural network
with 3-D Gauss-Hermite basis functions has been used. The neural network
contained 64 nodes in its hidden layer.