Professional Documents
Culture Documents
Bayesian Updating of Structural Models and Reliability Using Markov Chain Monte Carlo Simulation
Bayesian Updating of Structural Models and Reliability Using Markov Chain Monte Carlo Simulation
Abstract: In a full Bayesian probabilistic framework for ‘‘robust’’ system identification, structural response predictions and performance
reliability are updated using structural test data D by considering the predictions of a whole set of possible structural models that are
weighted by their updated probability. This involves integrating h()p(兩 D) over the whole parameter space, where is a parameter
vector defining each model within the set of possible models of the structure, h() is a model prediction of a response quantity of interest,
and p(兩 D) is the updated probability density for , which provides a measure of how plausible each model is given the data D. The
evaluation of this integral is difficult because the dimension of the parameter space is usually too large for direct numerical integration and
p(兩 D) is concentrated in a small region in the parameter space and only known up to a scaling constant. An adaptive Markov chain
Monte Carlo simulation approach is proposed to evaluate the desired integral that is based on the Metropolis-Hastings algorithm and a
concept similar to simulated annealing. By carrying out a series of Markov chain simulations with limiting stationary distributions equal
to a sequence of intermediate probability densities that converge on p(兩 D), the region of concentration of p(兩 D) is gradually portrayed.
The Markov chain samples are used to estimate the desired integral by statistical averaging. The method is illustrated using simulated
dynamic test data to update the robust response variance and reliability of a moment-resisting frame for two cases: one where the model
is only locally identifiable based on the data and the other where it is unidentifiable.
DOI: 10.1061/共ASCE兲0733-9399共2002兲128:4共380兲
CE Database keywords: Structural models; Markov chain; Reliability; Bayesian analysis; Monte Carlo method; Robust; Simulation.
冕
late samples according to an arbitrary PDF where the target PDF
need only be known up to a scaling constant. It was originally p 共 k⫹1 兲 ⫽ p 共 k 兩 k⫹1 兲 p D共 k⫹1 兲 dk
developed by Metropolis and his co-workers for computing ca-
nonical ensembles in statistical physics 共Metropolis et al. 1953兲
and later generalized by Hastings in Bayesian statistics 共Hastings
1970兲. Its potential use for solving reliability problems has been
⫽p D共 k⫹1 兲 冕 p 共 k 兩 k⫹1 兲 dk ⫽p D共 k⫹1 兲 (8)
recently demonstrated by Au and Beck 共1999兲. since 兰 p(k 兩 k⫹1 )dk ⫽1. This means that if the current sample
In the MH method, samples are simulated as the states of a k is distributed as the target PDF p D , then so is the next sample
special Markov chain whose limiting stationary distribution is k⫹1 , and hence p D is the stationary PDF of the Markov chain.
can populate sufficiently well the region of significant probability such as a symmetric one in the original Metropolis algorithm. If
of the target PDF p D . See Au and Beck 共1999兲 for a more de- the Markov chain is started in a region not near N, then the
tailed discussion of ergodicity in applying the MH method to chance of generating a candidate state from p * that visits N is
reliability problems. extremely small, and most of the candidate states will be rejected.
As a result, most of the Markov chain samples are repeated, and
Difficulties in Applying Metropolis-Hastings clearly they cannot be used to estimate R D . On the other hand, if
Algorithm to Reliability Updating the Markov chain is started in N, then in order for the simulated
candidate state to remain in N and have a high probability of
Using the Markov chain samples 兵 1 ,...,N 其 generated from the being accepted, the spread of p * around the current sample has to
MH procedure, R D is estimated as the average R̃ D of h over the be O(⑀), because the direction along which N extends around the
samples, which is the same as the usual MCS estimator, except current sample is not known. But this means the candidate state
that the samples are simulated from a Markov chain instead of will be very close to the current sample, and as a result the Mar-
being independent and identically distributed 共i.i.d.兲. Neverthe- kov chain samples will not efficiently explore N. In all these
less, the estimator R̃ D has similar statistical properties to those of cases, the region visited by the Markov chain samples will be
MCS estimators 共see later兲. In order to reduce the initial transient small compared to N, leading to significant bias in the estimate
effect of the Markov chain on the estimate, the first few samples for R D .
共say 10兲 are often not used to compute the estimate R̃ D . In this
work, the Markov chain samples 兵 1 ,...,N 其 used for computing Proposed Adaptive Simulation Method
the estimate R̃ D are those simulated after the initial transient
The problems encountered in applying the simulation methods
stage.
discussed in the previous sections, including the MH algorithm,
The proposal PDF p * in the MH algorithm affects the distri-
arise from the fact that the updated PDF p D is concentrated in a
bution of the candidate state given the current state, and conse-
small neighborhood N of the manifold S containing the maxima
quently the convergence rate of the estimator R̃ D to R D . If the of p D . The process of adapting samples to N in the MH algo-
candidate state is rejected too often because small values of r in rithm is inhibited by the small scale ⑀ of the thickness of N
Eq. 共4兲 are encountered during simulation, the Markov chain will compared to the size of the proposal PDF required to cover N
consist of many repeated samples. As a result, the correlation with an affordable number of samples. This suggests that direct
among samples will be increased, slowing down the convergence adaptation using a proposal PDF which varies with a vastly dif-
of R̃ D . To understand how p * affects the acceptance rate of the ferent length scale from that of the target PDF will not be effec-
candidate state, first note that, if p * (兩 )⫽p D(), then according tive. In view of this, we introduce a sequence of intermediate
to Eq. 共4兲, r⫽1 and hence the candidate state is always accepted. PDFs which bridge the gap in length scale between the prior PDF
In this case, the MH algorithm reduces to a standard Monte Carlo p 0 and the target updated PDF p D . By successively using the MH
procedure with i.i.d. samples simulated from p D . Although this algorithm to utilize the information from the previous adapted
choice is not possible, it indicates that, if p * is chosen to be intermediate PDFs, the region populated by the Markov chain
‘‘nonadaptive,’’ i.e., p * (兩 )⫽p * (), then the closer the p * is to samples gradually evolves into the desired region N. Conceptu-
the target PDF, the better the acceptance rate of the candidate ally, this is similar to ‘‘simulated annealing’’ 共Fishman 1996兲.
state, and the faster the convergence. Let p 1 ,...,p m be a chosen sequence of PDFs converging to
Choosing a nonadaptive proposal PDF p * means that the in- p D(⫽p m ) so that their region of significant probability content
formation from the current sample is not used to explore the gradually diminishes to that of p D . For example, p i may be cho-
important region of significant probability density of the target sen as the updated PDF from Bayes’ theorem based on an increas-
PDF during simulation. When information about the important ing amount of data, i.e., p i ⫽p Di , where D1 傺¯傺Dm ⫽D. The
region is not available, constructing a nonadaptive p * so that the proposed method is described as follows. Starting with the prior
candidate state simulated from it will lie in N is similar to con- PDF p 0 as the proposal PDF, the MH algorithm is carried out to
structing an importance sampling density concentrated in N, and simulate samples 兵 (1) 1 ,...,N 其 with target PDF p 1 . A kernel
(1)
is thus very difficult. It is therefore more desirable to choose an sampling density 1 is constructed as a weighted sum of Gaussian
‘‘adaptive’’ p * which depends on the current sample. One popular PDFs centered among these samples to approximate p 1 共Silver-
choice is to have p * localized and symmetric, i.e., p * (兩 ) man 1986; Ang et al. 1992; Au and Beck 1999兲.
⫽p * (兩 ), which corresponds to the one used in the original
N
Metropolis algorithm 共Metropolis et al. 1953兲. In this case, p *
can be interpreted as a PDF localized at the current sample. The 1 共 兲 ⫽ 兺 w k 共 ;共k1 兲 ,Ck 兲
k⫽1
(9)
Markov chain simulation process can then be viewed as a ‘‘local
random walk’’ in which the region of probability concentration of where (;(1)
k ,Ck )⫽multidimensional Gaussian PDF evaluated
the target PDF is adaptively explored. at with mean (1)k and covariance matrix Ck ; and
that is, the less p i⫹1 is changed compared to p i , the higher the
where „ j ;(1) acceptance rate of the candidate state, and the better the result at
k ( j),Ck ( j, j)…⫽one-dimensional Gaussian PDF
evaluated at j with mean (1) each simulation level. However, a slow schedule requires more
k ( j) and variance Ck ( j, j);
(1) ( j)⫽jth component of the vector (1) simulation levels to reach the final level for the updated PDF, and
k k ; and Ck ( j, j)⫽jth diag-
onal entry of the covariance matrix Ck . Similarly, the kernel mar- hence more Markov chain samples to be simulated. The prudent
ginal cumulative distribution function 共CDF兲 of j can be readily choice of an adaptation schedule is a trade-off between the effi-
computed in terms of a weighted sum of Gaussian CDFs. More ciency of the MH algorithm at each simulation level and the total
importantly, independent samples distributed as 1 can be readily number of simulation levels required.
simulated that are approximately distributed as p 1 and so lie in The choice of the adaptation schedule can be made by inves-
the region of significant probability of p 1 . In particular, to simu- tigating the probabilistic properties of the ratio r in the MH algo-
late a sample from 1 in Eq. 共9兲, first draw an index I from the set rithm appearing in Eq. 共4兲, which governs the acceptance rate of
兵 1,...,N 其 with corresponding probabilities 兵 w 1 ,...,w N 其 , and then the candidate states. It is found that a good choice for 兵 p i 其 is
simulate the sample as a Gaussian vector with mean vector I(1) p i⫹1 ⬃p 2i 共up to a normalizing constant兲. In this case, the size of
and covariance matrix CI . the region of probability concentration for p i⫹1 is roughly 1/&
To proceed, 1 is used as the proposal PDF for simulating that of p i . When the data D consist of measured response time
Markov chain samples 兵 (2) 1 ,...,N 其 with target PDF p 2 . These
(2)
histories 共Beck and Katafygiotis 1998兲, this schedule can be
samples are then used to construct the kernel sampling density 2 achieved by choosing p i ⫽p Di , where the duration of time history
as in Eq. 共9兲, which gives an approximation to p 2 . In general, the
data is doubled in successive simulation levels from Di to Di⫹1 .
kernel sampling density i 共which approximates p i 兲 is con-
structed using the Markov chain samples at the ith simulation On the other hand, when the updated PDF with data D is of the
level, which is then used as the proposal PDF for simulating form p D()⫽c exp关⫺J()/2⑀ 2 兴 共Vanik et al. 2000兲, where J()
Markov chain samples for the next level with target PDF p i⫹1 . is a measure-of-fit function between data and model, and ⑀ is a
This is continued until the mth simulation level, where Markov measure of the size of the prediction error, then the sequence 兵 p i 其
chain samples for the target updated PDF p D⫽p m are simulated. can be obtained by p i ⫽c i exp关⫺J()/2⑀ 2i 兴 where ⑀ 2i ⫽2 m⫺i ⑀ 2 , i
Let R i be the expectation of the response quantity h() when ⫽1,...,m, with 2 m ⬇⑀ ⫺2 , if the length scale of the prior PDF
is distributed as p i , that is p 0 () is O(1).
R i⫽ 冕 h 共 兲 p i 共 兲 d (11)
Statistical Properties of Estimators
Since p i converges to p D as the simulation level i increases to m, The statistical properties of the estimator R̃ i in Eq. 共12兲 are pre-
R i converges to R D defined by Eq. 共2兲. Using the Markov chain sented in this section, assuming the Markov chain generated ac-
samples 兵 (i)
1 ,...,N 其 , R i is estimated at the ith simulation level
(i)
cording to the MH algorithm at each simulation level is ergodic.
by R̃ i where In spite of the fact that R̃ i is computed using dependent
samples from a Markov chain, it still has the usual properties of
N
1 MCS estimators using i.i.d. samples 共Doob 1953兲. For example,
R i ⬇R̃ i ⫽ 兺
N k⫽1
h 共 共ki 兲 兲 (12)
R̃ i converges to R i with probability 1 as N→⬁ 共strong law of
large numbers兲, and under similar conditions as those for Monte
The proposed strategy makes use of the fact that, if a nonadap- Carlo estimators, R̃ i is normally distributed as N→⬁ 共central
tive proposed PDF is similar in shape to the target PDF, then the limit theorem兲. If the Markov chain is started with the initial state
acceptance rate of the candidate state will not be small, and the (i)
1 distributed as the target PDF p i , then the Markov chain is
MH algorithm will be effective in generating samples populating stationary, and R̃ i is unbiased, that is, E 关 R̃ i 兴 ⫽R i . Otherwise, R̃ i
the important region of the target PDF. Thus, instead of using an is only asymptotically unbiased, although the bias decays expo-
adaptive proposal PDF within each simulation level, the adapta- nentially with the number of Markov steps.
tion is done from one simulation level to the next, where the latest
An expression for the coefficient of variation 共COV兲 of R̃ i ,
kernel sampling density transfers the information about the im-
portant region from the current level to the next level. At each defined as the ratio of the standard deviation to the mean of R̃ i , is
simulation level, since the candidate state is simulated indepen- next derived assuming that the Markov chain has settled into its
dently of the current sample, successive Markov chain samples stationary state and the proposal PDF for the ith simulation level
are correlated only if the candidate state is rejected. With a rea- is fixed. As a result of stationarity, E 关 R̃ i 兴 ⫽R i . The variance of R̃ i
sonable acceptance rate of the candidate states, this leads to a is thus given by
关 h 共 共ki 兲 兲 ⫺R i 兴 册 2 that this additional variability is of the order of the bias in R̃ i , and
hence is often small. The numerical example shows that the COV
predicted by Eq. 共15兲 is quite close to the actual COV, and hence
N N it can be used for assessing the variability of R̃ i .
1
⫽ 2
N j⫽1 兺 k⫽1
兺 E 兵 共 h 共 共j i 兲 兲 ⫺R i 兲关 h 共 共ki 兲 兲 ⫺R i 兴 其
Illustrative Example
N N
1
⫽ 2
N j⫽1 兺 k⫽1
兺 c i共 j,k 兲 (13) The proposed methodology is applied to updating the response
variance and reliability of a structure using its identified natural
where c i ( j,k)⫽covariance between h((i) (i)
j ) and h(k ). Since the
frequencies for two cases: locally identifiable and unidentifiable.
Downloaded from ascelibrary.org by "Indian Institute of Science, Bangalore" on 09/18/21. Copyright ASCE. For personal use only; all rights reserved.
Markov chain is stationary and c i ( j,k) is symmetric in j and k, The actual structure is a two-dimensional finite-element model of
c i ( j,k) depends only on the absolute difference between the in- a two-story one-bay moment-resisting frame with a bay width of
dexes j and k, that is, c i ( j,k)⫽c i ( 兩 j⫺k 兩 ). Using this result and 8 m and equal story heights of 3 m. The columns have cross-
carrying out the double sum in Eq. 共13兲 with respect to ( j⫺k), sectional area of 18.8⫻10⫺3 m2 and moment of inertia 0.167
we have ⫻10⫺3 m4 . The beams have cross-sectional area of 10.5
⫻10⫺3 m2 and moment of inertia 0.562⫻10⫺3 m4 . The elastic
E 关 R̃ i ⫺R i 兴 2 ⫽
1
N2 冋
Nc i 共 0 兲 ⫹2
k⫽1
N⫺1
兺
共 N⫺k 兲 c i 共 k 兲 册 modulus and mass density of steel are assumed to be 2
⫻1011 N/m2 and 7,850 kg/m3, respectively. Lumped masses of
15⫻103 kg are put on the first and second floors. The natural
冋 冉 冊 册
N⫺1 frequencies of the first two modes are computed to be 3.16 and
c i共 0 兲 k
⫽
N
1⫹2
k⫽1
兺 1⫺ 共k兲
N i
(14) 9.67 Hz, respectively. Using noisy simulated response time histo-
ries, the identified natural frequencies are f̃ 1 ⫽3.13 Hz and f̃ 2
⫽9.83 Hz, which are used as the data D in the updating.
where i (k)⫽c i (k)/c i (0)⫽correlation coefficient at lag k, k
⫽1,...,N⫺1. The COV of R̃ i , denoted by ␦ i , is then given by
Locally Identifiable Case: Uncertain Stiffness Param-
E 关 R̃ i ⫺R i 兴 2 ⌬ i2 eters
␦ i2 ⫽ ⫽ 共 1⫹␥ i 兲 (15)
R i2 N A two-degree-of-freedom shear building model is used to model
the moment-resisting frame structure in order to identify the in-
where ⌬ i ⫽ 冑c i (0)/R i is the COV of h() when is distributed as terstory stiffnesses and to predict the response of the structure.
p i (), and ␥ i is a correlation factor The story masses are assumed to be 16.5⫻103 and 16.1⫻103 kg
冉 冊
N⫺1 for the first and second stories, respectively, which are computed
k
␥ i ⫽2 兺
k⫽1
1⫺ 共k兲
N i
(16)
based on lumping half of the mass of the columns on the floors
they are connected to. The interstory stiffnesses are parameterized
as k 1 ⫽k̄ 1 1 and k 2 ⫽k̄ 2 2 , where 1 and 2 are the stiffness
By estimating the covariance sequence 兵 c i (k):k⫽0,...,N⫺1 其
parameters to be identified, and k̄ 1 ⫽k̄ 2 ⫽29.7⫻106 N/m are the
from the Markov chain samples 兵 (i)
k :k⫽1,...,N 其 nominal values for the interstory stiffnesses of the first and second
N⫺k stories, respectively. The nominal values are computed assuming
1
c i共 k 兲 ⬇
N⫺k 兺
j⫽1
关 h 共 共j i 兲 兲 ⫺R̃ i 兴关 h 共 共j⫹k
i兲
兲 ⫺R̃ i 兴 (17) the beams are rigid 共shear building assumption兲. They do not
correspond to the interstory stiffnesses of the actual structure,
however, due to the flexibility of its beams.
the COV ⌬ i and the correlation sequence 兵 i (k):k⫽1,...,N⫺1 其 The prior PDF p 0 for 1 and 2 is given by the product of two
can be estimated. Consequently, ␥ i in Eq. 共16兲 and hence ␦ i in Eq. lognormal PDFs with most probable values 共MPVs兲 1.3 and 0.8
共15兲 can be estimated, providing a means for assessing the vari- for 1 and 2 , respectively, and unit standard deviation for both
ability of the estimate R̃ i using information from a single simula- 1 and 2 . These MPVs for the prior PDF reflect a 30% overes-
tion run. timation of the first-story stiffness and a 20% underestimation of
The term ⌬ 2i /N in Eq. 共15兲 is the familiar term for the square the second-story stiffness relative to the nominal shear building
of the COV in MCS with N independent samples. The COV of R̃ i model. The standard deviations for the prior PDF are chosen rela-
is thus equivalent to the COV in MCS with an effective number tively large to express large initial uncertainty in the structural
of independent samples N/(1⫹␥ i ). The efficiency of the estima- model.
tor using correlated samples of a Markov chain (␥ i ⬎0) is re- Using the modal data D, the updated PDF for the stiffness
duced compared to the case when the samples are uncorrelated parameters ⫽ 关 1 , 2 兴 is formulated as 共Vanik et al. 2000兲
(␥ i ⫽0), and smaller values of ␥ i imply higher efficiency.
The result for ␦ i in Eq. 共15兲 is derived assuming that the pro- p D共 兲 ⫽c exp关 ⫺J 共 兲 /2⑀ 2 兴 p 0 共 兲 (18)
posal PDF is fixed in independent simulation runs. According to where c⫽normalizing constant; and J()⫽modal measure-of-fit
the proposed methodology, however, the proposal PDF is chosen function
as the kernel sampling density i constructed using the Markov
chain samples from the previous simulation level 共except for the 2
first simulation level where the prior PDF is used as the proposal J 共 兲 ⫽ 兺 2j 关 f 2j 共 兲 / f̃ 2j ⫺1 兴 2 (19)
PDF兲, and so i is different for each independent simulation run. j⫽1
This gives rise to additional variability in R̃ i , and the actual COV In Eqs. 共18兲 and 共19兲, ⑀/ j is the coefficient of variation of the
of R̃ i will be greater than that given by Eq. 共15兲. It can be argued prediction error between each f̃ 2j and the corresponding model
2y 共 D兲 ⫽
j
冕 2y 共 兲 p D共 兲 d
j
(20)
冕
Fig. 3. Prior cumulative distribution function and updated marginal
cumulative distribution functions for 1 and 2 at simulation levels P 共 F j 兩 D兲 ⫽ P 共 F j 兩 兲 p D共 兲 d (21)
i⫽1, 5, 9
where P(F j 兩 ) is the first excursion probability for given model
parameters . Assuming the out-crossing events follow a Poisson
ginal PDFs and CDFs obtained by numerical integration are also process, P(F j 兩 ) can be approximated by 共Soong and Grigoriu
plotted with solid lines in Figs. 2 and 3, respectively. The results 1993兲
for the PDFs can be considered acceptable if fine detail is not
required. The results for the CDFs shown in Fig. 3 exhibit a better P 共 F j 兩 D兲 ⫽1⫺exp关 ⫺2 j 共 兲 T 兴 (22)
match with the exact results, as the spurious noise in the PDFs is where j () is the up-crossing rate for given model parameters ,
filtered out by integration in the CDFs. The errors in the marginal given by Rice’s formula
PDFs and CDFs come from two sources. The first source is that
the finite number of Markov chain samples is not distributed ex-
actly as the target PDF, which introduces bias in the estimates.
The second is due to the approximate nature of the kernel PDF
j 共 兲 ⫽
y j 共 兲
2 y j 共 兲
exp ⫺冉 b2
2 y j 共 兲 2 冊 (23)
using a finite number of samples. Thus, even if the samples are The estimates for updated robust standard deviations ˜ y j (D)
exactly distributed like the target PDF, there will still be discrep- ( j⫽1,2) are shown in Fig. 4 and the estimates for the updated
ancies between the kernel density estimates and the exact results. robust failure probabilities P̃(F j 兩 D) ( j⫽1,2) are shown in Fig.
The dependent nature of the Markov chain samples in general 5. In these figures, three sample estimates, corresponding to the
does not bias the kernel PDF, but it does slow down the conver- same simulation runs as in Figs. 2 and 3, are shown with circles.
gence of the kernel PDF compared to the case when the samples The exact results obtained by numerical integration are shown
are independent. with solid lines. Simulation level i⫽0 refers to the case when no
data are available and the robust response quantities are computed
Updated Robust Variance and Reliability based on the prior PDF p 0 only, that is, by Eq. 共2兲 with p 0 re-
The response of the structure when it is subjected to stochastic placing p D . To investigate the bias of the simulation results, the
excitations is predicted using the shear building model. Of inter- average of the estimates over 50 independent simulation runs is
est are the interstory drift responses y 1 and y 2 of the first and computed and shown with dashed lines in the figures. In both
second stories, respectively. The structure is assumed to be sub- Figs. 4 and 5, the results from averaging 50 simulation runs
jected to earthquake motion modeled by stationary Gaussian 共dashed lines兲 almost overlap with the exact results 共solid lines兲,
white noise with spectral intensity S⫽1⫻10⫺2 m2 /s3 . The sta- showing that the bias from the initial transient stage of the simu-
tionary variances 2y 1 () and 2y 2 () of the interstory drifts for lated Markov chains is negligible, and the estimates are practi-
given model parameters can be obtained by solving the cally unbiased.
Fig. 5. Estimates P̃(F j 兩 D), j⫽1, 2, for updated robust failure prob-
Fig. 7. Coefficient of variation of updated robust failure probability
abilities 共uncertain stiffness兲
estimates 共uncertain stiffness兲
Variability of Estimates
The sample COV of the updated robust drift variance and failure there is an infinite set of ‘‘optimal’’ mass and stiffness parameter
probability estimates computed from the 50 independent runs are values that give the two identified frequencies.
shown in Figs. 6 and 7. Also, the COV estimates based on Eq. The prior PDF p 0 for 1 to 4 is given by the product of four
共15兲, which does not include the variability of the proposal PDF, lognormal PDFs with MPVs 1.3,0.8,0.95,0.95, and standard de-
are computed and shown in these figures: the COV estimates for viations 1,1,0.1,0.1, respectively. Relatively small values of the
the three simulation runs in Figs. 2 and 3 are shown with circles, standard deviations are assumed for the mass parameters to reflect
while the averages of the COV estimates based on Eq. 共15兲 over that, in practice, they can usually be determined more accurately
50 simulation runs are shown with dashed lines. From Figs. 6 and from the structural drawings than the stiffness parameters.
7, it can be seen that the actual COV of the robust drift variance Fig. 8 shows the Markov chain samples at simulation levels
and failure probability estimates 共solid lines兲 are quite close to i⫽1, 5, 9 corresponding to ⑀⫽1, 1/4, 1/16 in Eq. 共18兲. At each
those predicted by Eq. 共15兲 共circles and dashed lines兲, indicating level, N⫽500 Markov chain samples are simulated. The four
that the additional variability due to the randomness of the pro- components of the samples are shown in two groups: 2 versus 1
posal PDF is negligible and Eq. 共15兲 is useful for assessing the in the first column and 4 versus 3 in the second column of Fig.
variability of the updated robust estimates. 8. The distribution of the samples for the stiffness parameters as
shown in the first column of the figure is qualitatively similar to
that shown in Fig. 1 but with more spread. There is no significant
Unidentifiable Case: Uncertain Stiffness and Mass Pa- pattern in the distribution of the mass parameters as shown in the
rameters second column of Fig. 8, as a result of the small uncertainty
Next, the story masses are also updated, in addition to the stiff- assumed in their prior distribution. In fact, the prior distributions
nesses. The story masses are parametrized as m 1 ⫽ 3 m̄ 1 and m 2 on the mass and stiffness parameters serve as a ‘‘regularizer’’ for
⫽ 4 m̄ 2 , where 3 and 4 are the mass parameters to be updated, this unidentifiable problem 共in the sense of Tikhonov and Arsenin
and m̄ 1 ⫽16.5⫻103 kg and m̄ 2 ⫽16.1⫻103 kg are the nominal 1977兲.
values for the first- and second-story masses, respectively. There
are thus four parameters 共two stiffness parameters 1 and 2 , and
two mass parameters 3 and 4 兲 to be updated. In this example,
the class of models is unidentifiable based on the data because
Fig. 6. Coefficient of variation of updated robust drift variance esti- Fig. 8. Markov chain samples for 1 to 4 at simulation levels i
mates 共uncertain stiffness兲 ⫽1, 5, 9
Fig. 9. Prior probability distribution function and updated marginal probability distribution functions for 1 to 4 at simulation levels i⫽1, 5, 9
共solid lines: average over 50 runs; dashed lines: three runs兲
The marginal kernel PDFs for 1 to 4 constructed using the shown in Fig. 11 and their COV is shown in Fig. 12. These results
Markov chain samples are shown in Fig. 9. The corresponding demonstrate the robustness of the proposed method to problems
marginal kernel CDFs are shown in Fig. 10. Calculation of the of higher dimensions.
exact marginal PDFs and CDFs involves three-dimensional nu-
merical integration, and was not attempted. Instead, the sample
average over 50 simulation runs of the marginal kernel distribu- Conclusions
tions are computed and are shown as solid lines in Figs. 9 and 10.
Finally, the estimates for the updated failure probabilities of The usual practice in system identification is to use system data to
the interstory drifts for the first and second stories are computed identify one model from a set of possible models and then to use
using the Markov chain samples. The results for the estimates are this model for predicting system behavior. In contrast, the present
Fig. 10. Prior cumulative distribution function and updated marginal cumulative distribution functions for 1 to 4 at simulation levels i⫽1, 5,
9 共solid lines: average over 50 runs; dashed lines: three runs兲
Acknowledgments