Information Technology Reference
In-Depth Information
40
20
0
−20
−40
−70
−60
−50
−40
−30
−20
−10
0
Real part
20
10
0
−10
−20
−700
−600
−500
−400
−300
−200
−100
0
Real part
Fig. 22 Poles distribution of beam reduced switched system (order 24) with DRK-SLS method
Table 4 H
1
error, CPU-time and the tolerance of iterative DRK-SLS and Iterative SVDDRK-
SLS
Models
Methods
H
1
Error
CPU-time (s)
Tol
FOM 1006
Iterative DRK-SLS
6
:
207
10
9
127
10
3
FOM 1006
Iterative SVDDRK-SLS
1
:
877
10
9
110
10
9
Beam 348
Iterative DRK-SLS
1
:
545
10
8
30
5
10
3
Beam 348
Iterative SVDDRK-SLS
1
:
081
10
8
55
2
10
2
reduces systems using the two proposed methods are presented, also the variation of
the absolute error between the original system and reduces systems are given. The
Table
4
contains the H
1
error, the CPU-time and the tolerance of each model. From
these
figures and this table, note that the iterative SVDDRK-SLS has a better results
compare to IDRK-SLS. Figure
23
shows the largest singular values of the fre-
quency response of the original switched linear system of order (1006) and reduced
one of order (10) to a frequency range using the IDRK-SLS and ISVDRK-SLS
methods, note a good correlation between the responses obtained by the reduces
systems and the original system, to understand the ef
ciency of one relative to the
other in the process of comparing the curves of absolute error variation.
The Fig.
24
presents the variation of the absolute error between the original
system (FOM of order 1006) and reduces systems (of order 10), note that the best
result is obtained by the ISVDDRK-SLS method. Figure
23
presents the largest
singular values of the frequency response of the original switched linear system of
order (348) and reduced one of order (24) to a frequency range using the IDRK-
SLS and ISVDRK-SLS methods. A good correlation over the whole frequency
range between the responses is obtained by the reduced system and the original one
Search WWH ::
Custom Search