Beruflich Dokumente
Kultur Dokumente
or in the
Applications
Troubleshooting
Finite Element model updating
Structural Modification
Reduction of the Order of mathematical models
Response Prediction
Structural Damage Detection
Active Vibration Control
Modal analysis is based upon the fact that the vibration response of
a linear time-invariant (LTI) dynamic system can be expressed as the
linear combination of a set of simple harmonic motions called the
natural modes of vibration.
This concept is akin to the use of a Fourier combination of sine and
cosine waves to represent a complicated waveform.
source: Jimin He, Zhi-Fang Fu, “Modal Analysis”, Butterworth-Heinemann
Institute of Structural Engineering Identification Methods for Structural Systems
Modal Analysis Techniques
Basic Concepts
Mẍ + Cẋ + Kx = F
In steady state conditions (i.e. once the free response has died
down), we showed that one can establish the relationship between
the complex amplitude of the response X and the input F = F0 e iωt
as:
X = H(iω) · F
In addition, index k corresponds to the k-th mode and the poles of the
denominator are simply the roots of the characteristic polynomial:
q
λk , λ∗k 2
= −ωk ζk ± i ωk 1 − ζk
∗
or hij (t) = (k Aij e λk t +k A∗ij e λk t )
q
with λk = −ωr ζk + iωk 1 − ζk2
Modal ID Fundamentals
then it follows that essentially, the FRF of a linear mdof system with
N dofs is the sum of N sdof FRFs (modal superposition) and the
transfer function matrix is completely characterized by the modal
parameters i.e the roots of the character polynomial λm and the
mode shape vectors Φm , m = 1, . . . , N
Given input and output time histories the FRF can be calculated through
the output/input Fourier Transform ratio:
U(ω)
H(ω) = Receptance FRF
F (ω)
U̇(ω)
H(ω) = Mobility FRF
F (ω)
Ü(ω)
H(ω) = Accelerance FRF
F (ω)
Syx (ω)
H(ω) = where
Sxx (ω)
Note: For the PSD to exist, the Random Process (signal) has to be
at least Wide Sense Stationary (WSS).
(source:http://cnx.org)
ACF
PSD
FT amplitude
plot(imp) 0.6
Impulse
xlabel('time') 1
ylabel('Impulse')
0.4
% Generate the Fourier 0.5
Transform 0.2
figure
plot(abs(fft(imp))) 0 0
xlabel('point') 0 20 40 60 80 100 0 20 40 60 80 100
ylabel('FT amplitude') time point
Power/frequency (dB/rad/sample)
x 10 Welch Power Spectral Density Estimate
MATAB code 4
% Generate 10000 point White 100
Noise 2
WN=wgn(1,10000, 100); 80
plot(WN)
WN
0 60
xlabel('time')
ylabel('WN') 40
figure -2
% Generate the Spectrum 20
pwelch(WN) -4
0
ylim([0 110]) 0 2000 4000 6000 8000 10000
0 0.2 0.4 0.6 0.8 1
time
Normalized Frequency (×π rad/sample)
FT amplitude
8*pi/(2*pi)=4Hz 0.6
sin(t)
fs=100; %sampling freq. 0
t=[0:1/fs:10];
0.4
s=sin(8*pi*t);
plot(t,s) -0.5
xlabel('time t') 0.2
ylabel('sin(t)')
figure -1 0
0 2 4 6 8 0 10 10 20 30 40 50
%Plot the FFT of the signal time t Hz
L=length(s);
NFFT= 2^nextpow2(L); % Next power of 2 from length of x; %FFT points need to be larger than the length of the
signal
Y = fft(s,NFFT)/L;
f = fs/2*linspace(0,1,NFFT/2+1); %Horizontal axis frequency equivalent in Hz
figure(2)
plot(f,2*abs(Y(1:NFFT/2+1)));hold on %Plot only half
xlabel('Hz')
ylabel('FT amplitude')
R1 R2
Q
2
ω ω1 ωn ω2 ω
The latter is derived from the SDoF system TF, (see Lecture 3) written for
mode n:
An
H(ω) = 2 2
ωn − ω + 2iζn ωωn
hence, for ω = ωn
An
Q = |H(ωn )| = ⇒ An = 2Qζn ωn2
2ζn ωn2
(qk is a scaling factor and does not affect the estimate of the mode)
Then the j-th component of mode k can be obtained either from:
p
the point FRF: Hjj ⇒ φjk = k Ajj or
k Aij
the transfer FRF: Hij ⇒ φjk =
φik
Disadvantages
This method relies on the peak FRF value, which is very difficult to
measure accurately
Damping is estimated from half power points only. No other FRF data
points are used. The half power points have to be interpolated, as it
is unlikely that they are two of the measured data points
This method which is very similar to the peak picking method is used for
the identification of the modal characteristics of the structure, given only
ambient measurements.
Assume the standard input X (t) - output Y (t) relationship, based on the
Fourier Transform:
Y (ω) = H(ω)X (ω)
When assuming wide sense stationary processes one can equivalently write
the following using PSDs:
Gyy (ω) = H ∗ (ω)Gxx (ω)H T (ω) (1)
where Gxx is the (r × r) power spectral density (PSD) matrix of the input,
r is the number of inputs, Gyy is the (m × m) PSD matrix of the
responses, m is the number of responses, H(ω) is the (m × r) frequency
response function (FRF) matrix and the overbar and superscript T denote
the complex conjugate and transpose, respectively.
Institute of Structural Engineering Identification Methods for Structural Systems
The FDD method
The method is applicable for cases where the Power Spectral Density
of the input is constant, i.e. Gxx (ω) = C . This applies for the case
of broadband random input excitations like white noise (such as
ambient excitations). In this way FDD is developed for output only
modal identification where only measurements of the response are
available.
Proof
µw (t) = E {w (t)} = 0
Rww (t1 , t2 ) = E {w (t1 ) w (t2 )} = C δ (t1 − t2 )
i.e. it is a zero mean process for all time and has infinite power at
zero time shift since its autocorrelation function is the Dirac delta
function. The above autocorrelation function implies the following
power spectral density, since the FT of Dirac Delta is unity:
Gww (iω) = C
Suppose the input y (t) is white noise, then the PSD is a constant matrix
⇒ Gxx (iω) = C . Hence,
N X
N
A∗m
X Am
(??) ⇒ Gyy (iw ) = + C
m=1 k=1
iω − λm iω − λ∗m
H
A∗k
Ak
+
iω − λk iω − λ∗k
where H
denotes the conjugate transpose (Hermitian transpose) matrix (∗T ).
Institute of Structural Engineering Identification Methods for Structural Systems
The FDD method
After some math manipulations, this can be further simplified to:
N
X Am A∗m Bm ∗
Bm
Gyy (ω) = + + +
iω − λm iω − λ∗m −iω − λm −iω − λ∗m
m=1
Therefore the contribution of only the mth mode in expression (2) is:
∗T
Rm CRm
Am =
−2Re(λm )
Dm ∝ Am CA∗T T T T
m = Φm γm C γm Φm = dm Φm Φm
Source: Brincker R., Zhang L., Palle A.,“Modal Identification from Ambient
Responses using Frequency Domain Decomposition”
Steps
A = UΣVH
where
U is a m × m unitary matrix: UHU = UUH = Im
Σ is a m × n diagonal matrix with Σii ≥ 0
VH is a n × n unitary matrix
Σii are called the singular values of A
(H denotes conjugate transpose)
Properties:
Gyy (ωi ) = Ui Σi UH
i
where Ui = [Ui1 , . . . , Uim ]
Isolate a peak
Singular Values
Frequency (Hz)
Assume two modal vectors φ, ψ. One expression for the MAC criterion is:
|ψ T φ|2
MAC =
(ψ T ψ)(φT φ)
for MAC = 1 ⇒ the vectors are consistent (similar)
MAC = 0 ⇒ not consistent
As long as the SVDs around that peak singular value give high MAC values
⇒ that singular vector belongs to the SDoF PSD.
Institute of Structural Engineering Identification Methods for Structural Systems
Identification Algorithm
Singular Values Isolate a peak – Equivalent sdof PSD Equivalent sdof time history
Acceleration
Tn
Frequency
Time
4 After we define the limits of the sdof, we can either apply the 1/2
power method to find ζ or we can take the signal back to the
frequency domain. There, we define fd (damped frequency) as 1/Td
from the zero crossings (or peak distance) and ζ from the logarithmic
decrement method.
5 In case of closely spaced modes, the first singular vector will always be
a good estimate of the strongest mode.
In the case though that these two closely spaced modes are
orthogonal, the first two singular vectors can give unbiased estimates
of the corresponding mode shapes.
Institute of Structural Engineering Identification Methods for Structural Systems
Identification Algorithm
Consideration
While the so-called mathematical poles, i.e. poles resulting from the
mathematical solution of the normal equations but meaningless with
respect to the physical interpretation, tend to vary in location.
(a)Model Order
Frequency vs
Institute of
Institute of Structural
Structural Engineering Identification Methods for Structural Systems