N=1000 st=0.02 SEC 
N=1000 At=0.02 SEC AR-10 
AR-5 = OPTIMUM 
o 
= 
@ 200 s 1.00 
= 2 
2 — 
> 
e g 
co 7) 
% 0.00 =1.00 
0.00 10.00 20.00 0.00 10.00 20.00 
FREQUENCY FREQUENCY 
N=1000 At=0.02 SEC 
N=1000 at=0.02 SEC 
eae AR-20 
o o 
o 
o 1.00 S00 
= 
= 5 
fea c 
BS S 
o w 
ie a 
wo n 
-1.00 -1.00 
0.00 10.00 20.00 0.00 10.00 20.00 
FREQUENCY FREQUENCY 
N=1000 At=0.02 SEC N=1000 At=0.02 SEC 
AR-30 AR-55 
2.00 2.00 
3 3 
= 
3 3 
= -0.00 = -0.00 
oO 12) 
a o 
B G 
-2.00 -2.00 
0.00 10.00 20.00 0.00 10.00 20.00 
FREQUENCY FREQUENCY 
Fig. 7.8. Effect of changing the orders in AR—model fitting (by MAIC; AR=10). 
(From Oda, Yamanouchi, et al.°°) 
indicates that the AR method gives better results than the B—T method and more precisely 
follows the rapid changes in the curves. 
By the way, the time shift techniques that this author proposed in Section 3.2 might 
have given even greater improvement in the results if it were properly applied. The rapid 
change in phase response with frequency shows this fact and indicates that there is still 
room for improvement by the proper shift of responses even in the case of using model 
fitting techniques. Figure 7.12** is an example of the same kind of comparison of two 
analysis methods used on the pitch motion of a model of an offshore semisubmersible as 
shown by Fig. 7.13 on irregular waves in the model tank. 
235 
