Structures
2.3 State-Space Structural Models
2.3.3 Modal Models
The ith component, or mode, has the state-space representation ( independently obtained from the state equations
, , )
mi mi mi
A B C
1
, , .
i mi i mi
i mi i
n i i
x A x B u y C x
y y
¦
(2.51)
This decomposition is justified by the block-diagonal form of the matrix and is illustrated in Fig. 2.9 for n= 2. In generic coordinates each state depends on itself (through the gain shown in Fig. 2.9 with a solid line) and on other states (through the gains shown in Fig. 2.9 with a dashed line). In modal coordinates the cross-coupling gains are zero, thus each state is independent and depends only on itself.
m, A Ami
Amij
Amij
y
y2
y1
u
x2
x1
x
2x1
Am21
Am12
Cm
Bm
Cm
Bm
³
³
+
+ +
+ + +
+
+
Am1
Am2
Figure 2.9. Block diagram of the state-space representation of a structure with two modes:
The modal cross coupling (marked with a dashed line) is nonexistent.
We consider three modal representations in this book. The blocks Ami, Bmi, and of these models are as follows:
Cmi
x Modal model 1:
0 0
, ,
2
mqi i
mi mi mi mvi
i i i mi i
A B C c
b Z
Z ] Z Z c ;
ª º
ª º ª º
« »
« » « »
¬ ¼ ¬ ¼ ¬ ¼
(2.52)
x Modal model 2:
, 0 , mqi
i i i
mi mi mi mvi i mvi
i i i mi i
A B C c c
b ] Z Z
Z ] Z Z ]
ª c º;
ª º ª º
« »
« » « »
¬ ¼ ¬ ¼ ¬ ¼
(2.53)
x Modal model 3:
2
0 1 0
, ,
mi 2 mi mi mqi mvi
i i i mi
A B C c
Z ] Z b
ª º ª º
c .
ª º
« » « » ¬ ¼
« » ¬ ¼
¬ ¼
(2.54)
Theith state component for the first modal model is as follows:
i mi ,
i mi
x q q
Z ½ ® ¾
¯ ¿ (2.55)
for the second modal model it is
i mi ,
i i i mi mi
x q
q q
Z ] Z
½
® ¾
¯ ¿ (2.56)
and for the third modal model it is
mi ,
i mi
x q q
½ ® ¾
¯ ¿ (2.57)
where and are the ith modal displacement and velocity, as defined in (2.18). Note that each component consists of modal displacement and velocity which, by (2.18), gives the original (nodal) displacement q and velocity Note also that eigenvalues of are the complex conjugate poles given by (2.32).
qmi qmi
q. Ami
We obtain the modal models 1, 2, and 3 from the corresponding state-space representations in modal coordinates as in (2.38), (2.42), and (2.45), by simply rearranging the columns of AandCand the rows ofA and B. Consider, for example, the third representation, with the state vector xT ªqTm qTmº,
¬ ¼ consisting of modal
displacements followed by modal rates. We transform it to the new state defined as follows:
1 1
1 2
2
2 ,
m m m m mn n mn
q
q x
q x
q x
q x q
½
° °
° ° ½
° ° ° °
° ° ° °
® ¾ ® ¾
° ° ° °
° ° °¯ °¿
° °
° °
¯ ¿
#
#
(2.58)
where the modal displacement for each mode stays next to its rate. The variable xi in the above equation is defined by (2.57). Formally, this state-space representation is derived using the transformation matrix R in the form
1 1
1 2
2 2
0 0
0 0
0 , where
0 0
0
i i
i n n
n
e
e E
e E e
e
R e
e E e
ª º
« »
« » ª º
« » « »
E ª º,
« » « » « »
« » « » ¬ ¼
« » « »
« » «¬ »¼
« »
« »
« »
¬ ¼
# # #
(2.59)
while is an n row vector with all elements equal to zero except the ith which is equal to one, and 0 denotes an n row vector of zeros (actually, we simply rearrange the coordinates).
ei
We obtain the modal models 1 and 2 in a similar manner by rearranging the states in the state vectors (2.36) and (2.40). The new state vectors for these representations are as follows:
1 1 1 1
1 1 1 1
1 1
2 2 2 2
2 2
2 , 2 2 2 .
m m
m m
m m
m m
n n
n mn n mn
mn n n mn mn
q q
q q 1
2 m
m
x q x
q q
x x
q q
x x q
x x
q q
q q
Z Z
] Z
Z Z
] Z
Z Z
] Z
½ ½
° ° ° °
° ° ½ ° ° ½
° ° ° ° ° ° ° °
° ° ° ° ° ° ° °
® ¾ ® ¾ ® ¾ ®
° ° ° ° ° ° °
° ° °¯ °¿ ° ° °¯ ¿
° ° ° °
° ° ° °
¯ ¿ ¯ ¿
# #
# #
q
¾°
°
(2.60)
In the above equation xi for modal models 1 and 2 are defined by (2.55).
We obtain the state representations from the representations as in (2.38) and (2.42), respectively, by rearranging the columns of AandCand the rows of A and B.
Note that the modal models (2.52), (2.53), and (2.54) are not unique in the sense that for the same matrix Ai, one can obtain different matrices Bi and , as explained in Appendix A.1. In particular, the first component of the input matrix
Ci
Bi
might not necessarily be zero, unless one finds a unique transformation that preserves the zero entry. For details see Appendix A.1.
In Appendix A the Matlab functions modal1m.m andmodal1n.m determine the modal model 1 using the modal or nodal data; and functions modal2m.m and modal2n.m determine the modal model 2 using the modal or nodal data. Also, the functionsmodal1.mandmodal2.m determine modal model 1 or 2 using an arbitrary state-space representation.
Example 2.9. Obtain modal model 2 from the model in Example 2.8.
Using transformation (2.59) we find
0.0487 3.1210 0 0 0 0
3.1203 0.0487 0 0 0 0
0 0 0.0233 2.1598 0 0
0 0 2.1596 0.0233 0 0 ,
0 0 0 0 0.0030 0.7708
0 0 0 0 0.7708 0.0030
Am
ª º
« »
« »
« »
« »
« »
« »
« »
« »
¬ ¼
1 2 3
0 0.3280 0 0.5910 0 0.7370
m
m m
m
b
B b
b
ª º
« »
« » ª º
« » « »
« » « »
« » «¬ »¼
« »
« »
« »
¬ ¼
,
and
>
1 2 3@
0.1894 0 0.3412 0 0.4255 0
0.0092 0.5910 0.0080 0.7370 0.0013 0.3280 .
0.0051 0.3280 0.0064 0.5910 0.0028 0.7370
m m m m
C c
ª º
« »
« »
« »
¬ ¼
c c
The transfer function of a structure is defined in (2.5). In modal coordinates it is, of course,
(2.61) ( ) m( m) 1 .
G s C sI A Bm
As was said before, the transfer function is invariant under the coordinate transformation; however, its internal structure is different from the generic transfer function (2.6). In modal coordinates the matrix sI A m is block-diagonal, and it can be decomposed into a sum of transfer functions for each mode, therefore,
Property 2.2. Transfer Function in Modal Coordinates. The structural transfer function is a composition of modal transfer functions:
2 2
1 1
( )
( ) ( ) ,
2
n n
mqi mvi mi
mi
i i
i i
c j c b
G G
j
Z Z Z
Z Z ] Z Zi
¦ ¦
(2.62)where
1
2 2
( )
( ) ( ) , 1, , ,
2
mqi mvi mi
mi mi mi mi
i i i
c j c b
G C j I A B i
j
Z Z Z
Z Z ] Z Z
!n (2.63)
is the transfer function of the ith mode. The value of the transfer function at the ith resonant frequency is approximately equal to the value of the ith mode transfer function at this frequency:
2
( )
( ) ( ) .
2
mqi i mvi mi
i mi i
i i
jc c b
G G Z
Z Z
] Z
# (2.64)
Proof. IntroducingA, B, andC as in (2.45) to the definition of the transfer function we obtain
1 1
1 1
( ) ( ) ( ) ( ),
n n
m m m mi mi mi mi
i i
GZ C j I AZ B
¦
C j I AZ B¦
G Zwhich proves the first part. The second part follows from the first part by noting that, for flexible structures with distinct natural frequencies and low damping,
2 2
( ) ( )
mj i mi i
G Z G Z for iz j.
Special Models
ª how to describe less-common structures
Do not quench your inspiration and your imagination;
do not become the slave of your model.
—Vincent van Gogh
Models described in the previous chapter include typical structural models, which are continuous-time, stable, and with proportional damping. In this chapter we consider models that are not typical in the above sense but, nevertheless, often used in engineering practice. Thus, we will consider models with rigid-body modes (which are unstable), models with nonproportional damping, discrete-time structural models, models with acceleration measurements, and generalized structural models.
The latter include two kinds of inputs: controlled (or test) inputs and disturbance inputs, and also two kinds of outputs: measured outputs and outputs where the system performance is evaluated.