Accepted manuscript doi:

10.1680/jstbu.18.00121
Accepted manuscript
As a service to our authors and readers, we are putting peer-reviewed accepted manuscripts
(AM) online, in the Ahead of Print section of each journal web page, shortly after acceptance.

Disclaimer
The AM is yet to be copyedited and formatted in journal house style but can still be read and
referenced by quoting its unique reference number, the digital object identifier (DOI). Once
the AM has been typeset, an ‘uncorrected proof’ PDF will replace the ‘accepted manuscript’
PDF. These formatted articles may still be corrected by the authors. During the Production
process, errors may be discovered which could affect the content, and all legal disclaimers
that apply to the journal relate to these versions also.

Version of record
The final edited article will be published in PDF and HTML and will contain all author
corrections and is considered the version of record. Authors wishing to reference an article
published Ahead of Print should quote its DOI. When an issue becomes available, queuing
Ahead of Print articles will move to that issue’s Table of Contents. When the article is
published in a journal issue, the full reference should be cited in addition to the DOI.

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121
Submitted: 08 July 2018
Published online in ‘accepted manuscript’ format: 18 February 2019
Manuscript title: Structural damage detection based on modal data using moth-flame
optimization algorithm
Authors: Mohamad Mohamadi Dehcheshmeh1, Gholamreza Ghodrati Amiri2, Ali Zare
Hosseinzadeh2 and Vahid Torbatinejad1
Affiliations: 1School of Civil Engineering, Iran University of Science and Technology,
Narmak, Tehran, Iran and 2Center of Excellence for Fundamental Studies in Structural
Engineering, School of Civil Engineering, Iran University of Science & Technology, Tehran,
Iran
Corresponding author: Gholamreza Ghodrati Amiri, Center of Excellence for Fundamental
Studies in Structural Engineering, School of Civil Engineering, Iran University of Science &
Technology, Tehran, Iran.
E-mail: [email protected]

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121
Abstract

This article presents a new and efficient damage detection method for civil structures. To identify damage

localization and quantification, the proposed method is defining damage detection problem as an inverse

problem. The objective function is based on modal strain energy (MSE) and static displacements. A recently

developed Nature-inspired Heuristic Paradigm called Moth-Flame Optimization (MFO) algorithm by mirjalili

(2016) is utilized and has been compared with other algorithms like PSO (particle swarm optimization), and CS

(cuckoo search). for simulating the real condition for SHM programs, challenges like contaminating frequency

with random noise and using restricted mode shape data are also taken into account. In order to check the

robustness of the proposed method, several examples, including a plane steel truss, plane steel frame, and eight-

storey shear frame are considered. Also several damage scenarios have been considered for each example.

Errors between simulated and expected result in each damage scenario were less than 2 percent in numerical

studies showing that the efficiency and capability of present damage detection method is acceptable.

Keywords: damage detection; health and safety; steel structures; modal strain energy; static displacements;

seismic engineering; moth-flame optimization algorithm

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121
Introduction

As the time goes by, more civil structures are getting to their critical age. Correct and early
detection of the probable damage prevents catastrophic affairs. In the last few decades
scientists tried to gain reliable, non-destructive and efficient damage identification technique.
Vibration-based structural health monitoring (SHM) is one option to inspect structure's
probable flaw and because of their efficiency, have gained researcher's attention recently.
Among the vibration-based methods, those based on modal analysis are extensively used, as
the modal parameters depend on only mechanical characteristics of a structure and not the
excitation applied. Lifshitz et al. (1969) proposed utilizing the shifts in natural frequencies to
discover structural failure. Since then, a great number of researches about damage detection
considering the changes in frequencies have been published, Rytter (1993) investigated
dynamic behaviour of offshore structures and numerical and experimental studies were
considered. Salawu (1997) has carried out a general review of this research. Most of the
methods are based on finding of characteristics patterns of change in the natural frequency of
structure or on observation of differences in shape of natural modes and static deflections as
well as on accumulated strain energy. Sohn et al. (2003) produced a summary of the
structural health monitoring literature in 1995. This presentation will summarize the outcome
of an updated review covering the years 1996 - 2001. Yan et al. (2007), Fan et al. (2011) and
Carden and Fanning (2004) also presented comprehensive review of these investigations.
Numerous vibration-based methods are defining damage identification problem as an inverse
problem and solving it using optimization algorithms (Du et al. 2017; Saada et al. 2013; Hao
et al. 2002; Chou et al. 2001; Teughels et al. 2005; Betti et al. 2015; Sandesh et al. 2010;
Bagheri et al. 2012; Nanda et al. 2014; Khatir and Abdel Wahab 2018; Kang et al. 2012;
Ding et al. 2017; kaveh 2017; Samir 2017; kaveh and zolghadr 2017; khatir 2018). a group of
studies use modal strain energy as damage index in SHM (Shi et al. 1999; Li et al. 2006;
Sazonov, and Klinkhachorn, 2005; Hu,et al. 2006; Shi, et al. 1998; Kim et al. 2003; Yang et
al. 2016) Perera et al. (2009) proposed single and multi-objective functions for damage
detection by using changes in both natural frequencies and mode shapes, and found optimal
solution by utilizing Genetic Algorithm. Seyedpoor (2012) introduced a two stage method
defining a cost function utilizing modal strain energy and optimized it via particle swarm
optimization algorithm and tested the efficiency on truss and beam. Zare Hosseinzadeh et al.
(2016) presented a novel method to detect and estimate structural damage based on flexibility
matrix and used democratic particle swarm optimization (DPSO) algorithm to minimize the
objective function. The method was tested in a condition in which limited number of sensors
are installed on the structure. Zare Hosseinzadeh et al. (2016) also suggested a method for
detecting structural damage by employing Cuckoo Optimization Algorithm (COA). They
verified their method by studying different numerical and experimental examples. Khatir et
al. (2017) presented a method for inverse problem in order to detect and localize damage in
composite beam-like structures. The proposed technique makes use of Genetic Algorithm
(GA) and Particle Swarm Optimization (PSO) methods. Tiachacht et al. (2018) proposed a
new methodology for damage identification and quantification in two- and three-dimensional
structures using a Modified Cornwell Indicator combining with Genetic Algorithm (GA).
Kaveh, and Dadras (2018) utilized a recently developed optimization algorithm called
thermal exchange optimization algorithm (TEO) and applied to a damage detection problem.
Several scenarios with noise and noise-free modal data were tested and the location and
extents of damage are identified with good accuracy. Iacono, et al. (2012) validated the
feasibility of an identification procedure, based on Hilbert transform, by means of

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121
experimental tests for shear-type multi-degree-of- freedom. They tested the method on
experimental and numerical examples. Zhu et al. (2011) developed a damage detection
method for shear buildings using the changes in the first mode shape slopes. They used an
eight-storey shear building for numerical study and a three-storey shear building was
considered for the experimental validation. Abdeljaber et al. (2017) presented a novel, fast and
accurate structural damage detection system using 1dimentional Convolutional Neural Networks
(CNNs) that has an inherent adaptive design to fuse both feature extraction and classification blocks
into a single and compact learning body. The proposed method performs vibration-based damage
detection and localization of the damage in real-time.

They use all members of the calculated characteristic from modal data or use a large number
of modal data for creating damage-sensitive cost functions. Furthermore, most of them follow
a direct data-fitting strategy for introducing cost functions. Since using many recorded data
by a direct data-fitting approach may intensify the noise effects, in this article noises in been
applied in the cost function. In this article a new strategy is proposed for damage
identification in shear frames by calculating the static displacements and accumulated strain
energy in each of the stories using modal and optimizing via a recently developed optimizer
called the moth-flame optimization algorithm. The efficiency of presented method are
investigated by a numerical examples which is a plane steel truss, plane steel frame, and an
eight-storey shear frame, and challenges like applying direct noise to the data and the effect
of the number of available modal data is considered. In order to check the reliability of the
proposed algorithm, the proposed method is tested by two different optimization algorithms.

The rest of the article is sorted as follows. A general review of moth- flame optimization
algorithm, presenting a new damage identification method, introducing the numerical
examples and demonstrating obtained results and finally the article will be finished with
discussion and conclusions.

2. Moth-flame Optimization algorithm

The Moth-Flame algorithm is a recently developed meta-heuristic optimization algorithm that

mimics the navigation method of moths in the night. In this method, the moths are the
candidate solutions and the moths' positions are the problem's parameters. Also, moths can
fly in I-D, 2-D, 3-D, space by switching their position vectors. Mirjalili (2016)
mathematically model this behaviour and propose an optimizer called Moth-Flame
Optimization (MFO) algorithm.

To generate the population of MFO algorithm, n moths and flames occupy matrix M and F,
respectively. Both matrices have n  d dimension where n is the number of moths and flames
and d is the number of parameters. Matrix OM and OF sort the components according to the
value of objective function and are defined as follows (Mirjalili 2016):

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121
OM 1 
OM 
OM   2

 
 
OM n  (1)

OF1 
OF 
OF   2 
 
 
OFn  (2)

The moths are actual search agents that move around the search space, whereas flames are the
best position of moths that obtains so far.

With this mechanism, a moth never lose its best solution. The MFO algorithm contains
three-tuple that approaches the global optimum of the optimization problems and defined as
follows (Mirjalili 2016):

MFO  (I , P ,T ) (3)

I is a function which creates the population of moths randomly and fitness values of them,
The P function, which is the main function, is the important function that define how the
moths move around the search area. This function received the matrix of M and returns its
updated one finally. And the T function returns true if the termination criterion is satisfied
and false if the termination criterion is not satisfied. The function I has to generate initial
solutions and calculate the objective function values. In P function, update the location of
each moth respect to matching flame using Equations (4, 5, and 6) (Mirjalili 2016):

M i  S (M i , Fj )
(4)

M F
Where i indicate the i-th moth, j shows the j-th flame, and S is a logarithmic spiral
function for the MFO algorithm as follows (Mirjalili 2016):

S (M i , Fj )  Di .e bt .cos  2 t   Fj
(5)

D
Where i indicates the distance of the i-th moth for the j-th flame, b is a constant for
describing the shape of the logarithmic spiral, and t is a random number in [-1,1]. D is
considered as follows (Mirjalili 2016):

Di  Fj - M i
(6)

M F D
Where i shows the i-th moth, j indicates the j-th flame, and i point to the distance of
the i-th moth for the j-th flame. Spiral movement is the main component of the proposed
method because it dictates how the moths update their positions around flames.

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121
Mirjalili demonstrated that the MFO algorithm has the capability to discover very
competitive results compared with other famous meta-heuristic algorithms such Particle
Swarm Optimization (PSO) due to two reasons First, the diversification of MFO algorithm is
very high and needs it to avoid local optima. Second, the equilibrium of diversification and
intensification is very simple and effective in finding the optimal solution to solve real
problems (Mirjalili 2016). In this paper, MFO algorithm is used to optimize the objective
function and identify location and severity of damage in elements.

3. Proposed method

In this section a new damage index is proposed for structural damage detection, using
dynamic data of a structure including modal information and stiffness matrices. For a
structure, free vibration equation can be defined as:
Mx+Kx=0 (7)
Where M and K are mass and stiffness matrices, respectively. Also x and x are acceleration
and displacement vectors, respectively. Natural frequencies ( ωi ) and mode shapes ( φ i ) can
be extracted from equation below:
(K  M i2 )i  0 i=1,….,n (8)

Considered damage is defined as reduction in stiffness matrix of damaged elements. So the

analytical model for stiffness matrix in damaged structure for unknown damage severity is
written as:
Ked  Kue (1  d e ) (9)

Where Ked and Kue are the stiffness matrices of e-th element in damaged and undamaged
conditions, respectively. Also, d e is the damage severity of e-th element which is a number
between 0 and 1, zero represents an undamaged element and one represents complete
damaged element. In conclusion the stiffness matrix of the analytical model is defined as:
K (d 1 , d 2 , d 3 ,..., d Ne )  eNe1 K de (10)

It is assumed that a unique static load such as Fo is applied to a structure with n degrees of
freedom. This load is defined as (Zare Hosseinzadeh et al. 2014):
Fo  1 1 1 ... 1
T
(11)

The static equilibrium equation of a structure can be expressed as follows:

Fo  K  (12)

Where K and  are the stiffness matrix and displacement vector, respectively. The vector
static displacement of structure can be presented as:
  (k )1 Fo  GFo (13)

Where G is the flexibility matrix. The flexibility matrix for structure can be defined as (Zare
Hosseinzadeh et al. 2014):
G  1T (14)

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121
Where  is the normalized mode shape matrix and  is a diagonal matrix whose members
are the eigenvalues of free vibration problem as follows:
12 0 ... 0 
 
 0 22 ... 0 
 (15)
 ... ... ... ... 
 
 0 0 ... N2 

Normalized mode shape matrix is extracted from formula below:

T M   I (16)

In order to simplify computation, NMU i is a parameter which is extracted from static

displacement as follows:

i
NMU i  n
i=1,…,n (17)
 i
i

Where in NMU i i stands for i-th degree of freedom.

For damage structure and model the above parameter is defined as NMU d and NMU m ,
respectively. Usually, for two vectors of x and y, MAC can be taken to mean as a standard for
assessing amount of geometrical correlation between them which is defined as below (Zare
Hosseinzadeh et al. 2017):
2
x T .y
MAC (x , y )  (18)
(x T .x ).( y T .y )
Considering NMU d and NMU m as two vectors, MAC1 is defined as follows:
2
NMU d T .NMU m
MAC 1(x , y )  (19)
(NMU d T .NMU d ).(NMU mT .NMU m )
Strain energy is stored in each element of a structure. The modal strain energy (MSE) of j-th
element of a structure can be defined as (Seyedpoor 2012):
1
MSE j  Tj K j  j j=1,…..,n (20)
2

Where K j is the stiffness matrix of j-th element of a structure and  j is the vector of
corresponding nodal displacements of element j. By summation of MSE of all elements, the
total strain energy of the structure can be extracted as below:
nte
MSET   MSE j j=1,…..,nte (21)
j 1

Where nte is the number of structure's elements. For simplicity of computing, NMSE j is a
parameter extracted from MSE and defined for all elements as below:

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121
MSE j
NMSE j  (22)
MSET
Where in NMSE j j stands for j-th element of the structure. Generally damage occurrence is
shown by reduction in stiffness parameters of a structure, because of this, damage occurrence
is the reason for increasing MSE and consequently NMSE j for each element. For damage
structure and model the above parameter is defined as nmsed and nmse m , respectively.
Considering NMSE d and NMSE m as two vectors, MAC2 is defined as follows:
2
NMSE d T .NMSE m
MAC 2(x , y )  (23)
(NMSE d T .NMSE d ).(NMSE mT .NMSE m )

The damage detection problem can be stated as an optimization problem by manipulating

equation (19) with equation (23), F function as a cost function is defined as:
F (d1 , d 2 , d 3 ,..., d Ne )  abs ((1  MAC 1).(1  MAC 2)) (24)

The problem is solved by MFO algorithm by minimizing the cost function F. The changes in
masses are neglected and is assumed to be the same in healthy and damaged structure. to
evaluate this method, different examples, including several scenarios is anticipated and
different value of random noise are also considered.

4. Numerical studies
In this section the efficiency of the proposed method is tested. Challenges like random noise
is applied in each study. To check the robustness of the method, three different example
including a plane steel truss, a plane steel frame, and a shear frame has been considered. It
should be noted that the complete work has been handled with MATLAB workspace. It is
worth mentioning that the number of population in moth-flame algorithm for all examples
equals to 150. Moreover, the number of iterations for plane steel truss and plane steel frame
are 1000 and for shear frame is 200. It should be noted that these numbers are gained by trial
and error. All software runs of the optimization algorithm is done by CPU with follow
characteristics: intel® core™2 dou.

4.1 plane steel truss

In the first example, a plane steel truss is studied in order to identify damage location and
quantification. The finite element model of the plane steel truss is shown in Figure 1.

The material properties of this example are as follows: modulus of elasticity and mass density
for all elements are 200 GPa and 7850 kg/m3, respectively. The cross-sectional area and mass
per unit for vertical members are 0.005 m2, and m=39.25 kg/m, and those for bottom
horizontal members are 0.01 m2, and 4000 kg/m, and those for top horizontal members are
0.01 m2, and m=78.50 kg/m, and those for diagonal members are 0.008 m2, and m=62.80
kg/m, respectively. As we know, each free node of a planar truss has two DOFs,
consequently, this example has 12 DOFs.
Three damage scenarios is considered to test the proposed method, Table 1 elaborates these
scenarios.

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121
Damage severity in each scenarios are divided to light, normal and severe damage, the
number of damaged elements is different in each scenarios. In real structural health
monitoring (SHM), initial data is contaminated with some noises, for actual evaluation of the
proposed damage detection method, noise are applied to the initial data and being compared
with the state that noise is not included. Based on this strategy, natural frequencies are
contaminated with noise according to the formula below:
(25)
Where and are i-th natural frequency of the structure with and without the noise,
respectively. Also and are noise level and a random value between -1 and 1 which is
generate with Matlab software, respectively. To examine the noise effect on the method, 5
percent of noise is applied to the natural frequencies of the structure and also the actual
structure without noise is tested for comparison. Details of the results for different damage
scenarios are shown in figures 2-4. Also the program’s run time for different scenarios and
1000 iterations is shown in figure 5. Figure 5 indicates that time of run is not a subject of
different scenarios with several damage in the structures. Figure 6 shows the convergence
plot of MFO algorithm for scenario 3, using different number of modes with 0 and 5 percent
noise. To evaluate the applicability of the MFO algorithm, another optimization algorithm is
used to compare with the proposed method. Figure 7-8 show results of damage detection with
PSO and Cuckoo Search (CS) algorithms, it can be seen that these algorithms don't detect
damage location and quantification well enough. Comparison of Convergence plots between
MFO, PSO, and CS algorithms in plane steel truss for third scenario using one mode's data
figure 9. As it can be seen in figure 9, the MFO algorithm converge faster and in less number
of iteration in comparison with other considered algorithms. To investigate the performance
of three algorithms, time of run for MFO, PSO, and CS algorithms is compared in figure 10.
It is implied that PSO algorithm consumes approximately same time in comparison with the
proposed MFO algorithm, although the CS algorithm takes time about two times longer than
other two algorithms.

4.2 plane steel frame

An asymmetrical plane steel frame is the second example, of which, each free node of the
structure has three DOFs, the finite element model of the frame is shown in figure 11.

Material properties of the frame are as follows: the modulus of elasticity and mass density
are 200 GPa and 7850 kg/m3, respectively. The mass per unit length, moment of inertia, and
cross-sectional area are 133.45 k/m, 0.00040 m4, and 0.0170 m2 for columns, and 1200 k/m,
0.000435 m4, and 0.0172 m2 for beams, respectively.
Table 2 demonstrates the details of three damage scenarios that are considered for this
example.
Actual structure without noise and structure with 5 percent noise polluting the frequencies are
analysed. As it was mentioned before, the cost function is optimized by MFO algorithm, the
obtained results are shown in figures 12-14. As the structure got more complicated the time
of run of the optimization algorithm was longer, but as mentioned before, different scenarios
didn't have effect on the time, the result is shown in figure 15. Convergence plot of the MFO
algorithm for plane steel frame in scenario 3 with different modes’ data and different noise
percent is shown in figure 16.

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121
4.3 Eight-storey shear frame

In third example, the efficiency of proposed method is investigated by studying an example,

which is an eight-storey shear frame. In the example, damage is simulated as a relative
reduction in the stiffness of each storey or element. The finite element model of the structure
is shown in figure 17.

The physical properties of the numerical shear frame model is listed in Table 3.The severity
of damages as an output data is gained from MFO algorithm. Table 4 elaborates these
scenarios.

Damage severity in each scenario is divided to light, normal and severe damage. Moreover,
the number damaged elements is different in each scenario. 0, 5, and 7 percent noise are
applied to the initial data and is compared with the state that noise was not included. Figures
18-20 show results of the analysed model. The Structure in this example is less complicated
than the other examples, so as mentioned before, the time of run of the program was shorter,
as it can be seen in figure 21. Convergence plots are shown in figure 22 for scenario 3 with
different percent of noise and number of used modes.

5. Discussion
The aim of this paper is to identify and localize damaged elements in different structures. The
proposed objective function was tested by MFO algorithm and compared with PSO and CS
(cuckoo search) algorithm. Vital items for this study are detection of damage location,
damage severity, time of operation and convergence plot. As it was expected, the structures’
complexity affected on time of operation. For example, the simplest structure that has been
analysed was steel shear frame, the time of operation was about 10 seconds. However, plane
steel truss and plane steel frame were operated in 250 and 400 seconds, respectively. Also it
can be understood that time of operation is not a subject of different scenarios like one or four
elements damage scenarios. It is worth mentioning that despite the fact that PSO algorithm
consumed approximately same time in comparison with the proposed MFO algorithm and CS
algorithm takes time about two times longer than the other two algorithms, the power of the
MFO algorithm to detect and localize the damage is much more than the PSO, and CS
algorithm. It can be seen that convergence of three compared algorithms was different. For
example in the situation with no random noise, the MFO algorithm was converged to zero in
200 iterations. However, the PSO algorithm was converged by more than 1000 iterations and
also MFO converged faster than CS algorithm. To conclude, the proposed objective function
using modal strain energy and static displacement, is found capable to recognise damage in
the structure. Moreover, the optimization algorithm has been qualified with different
conditions, like 5 and 7 percent noise polluting the natural frequencies and by using different
modal information.

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121
6. Conclusion

In this article a new damage identification and quantification method was proposed. The
problem was defined as an inverse problem by means of using modal strain energy (MSE)
and static displacement. The moth-flame optimization algorithm was used to optimize the
objective function. In order to check the capability of the propose damage detection method,
several examples was tested. A thirteen-member plane steel truss with 12 degrees of freedom,
a plane steel frame with 27 degrees of freedom, and an eight-storey shear frame was
considered. Moreover, to simulate a real SHM program, challenges like applying different
quantity of random noise to natural frequencies, defining several damage scenarios for each
example, and using restricted number of mode shapes data, was considered. The proposed
optimization algorithm was compared with two well-known algorithms that is PSO and CS
optimization algorithms. The error between simulated and estimated result in each damage
scenario were less than 2 percent in this study. All results approved perfect performance of
the proposed method. Therefore, the presented method can be introduced as an efficient
method for structural damage identification in real SHM programs.

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121
List of notation
x Acceleration vector
x Displacement vector
M Mass matrix
K Stiffness matrix
ωi Natural frequencies
φi Mode shapes
e
K d Stiffness matrices of e-th element in damaged condition
e
K u Stiffness matrices of e-th element in undamaged condition
de Damage severity of e-th element
 Static displacement vector
G Flexibility matrix
 Normalized mode shape matrix
 Diagonal matrix whose members are the eigenvalues of free vibration
NMU i Parameter extracted from static displacement of i-th degree of freedom
NMU d Parameter extracted from of i-th degree of freedom for damaged structure
NMU m Parameter extracted from of i-th degree of freedom for model of structure
MAC geometrical correlation between two vectors
mse j Modal strain energy of j-th element
mseT Total strain energy of a structure
nmse j
Parameter extracted from MSE of j-th element of the structure
I function that generates a random population of moths and corresponding fitness values
P function which is the main function, moves the moths around the search space
Mi The i-th moth,
Fj
The j-th flame
S logarithmic spiral function for the MFO algorithm
Di
Indicates the distance of the i-th moth for the j-th flame
sort the components according to the value of objective function
OM
sort the components according to the value of objective function
OF
Noise level
A random value between -1 and 1

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121
References

ABDELJABER, OSAMA, ONUR AVCI, SERKAN KIRANYAZ, MONCEF GABBOUJ, and DANIEL J
INMAN. 2017. Real-time vibration-based structural damage detection using one-
dimensional convolutional neural networks.Journal of Sound and Vibration, 388:
154-70.
BAGHERI, A., RAZEGHI, H. & AMIRI, G. G. 2012. Detection and estimation of damage in
structures using imperialist competitive algorithm. Shock and Vibration, 19, 405-419.
BETTI, MiICHELE, LUCA FACCHINI, and PAOLO BIAGINI. 2015. Damage detection on a three-
storey steel frame using artificial neural networks and genetic algorithms.
Meccanica, 50: 875-86.
CARDEN, E PETER, and PAUL FANNING. 2004. Vibration based condition monitoring: a
review. Structural health monitoring, 3: 355-77.
CHOU, J.-H. & GHABOUSSI, J. 2001. Genetic algorithm in structural damage detection.
Computers & structures, 79, 1335-1353.
CORNWELL, P., DOEBLING, S. W. & FARRAR, C. R. 1999. Application of the strain energy
damage detection method to plate-like structures. Journal of sound and vibration,
224, 359-374.
DING, Z., YAO, R., HUANG, J., HUANG, M. & LU, Z. 2017. Structural damage detection based
on residual force vector and imperialist competitive algorithm. Structural
Engineering and Mechanics, 62, 709-717.
DU, D.-C., VINH, H.-H., TRUNG, V.-D., HONG QUYEN, N.-T. & TRUNG, N.-T. 2017. Efficiency of
Jaya algorithm for solving the optimization-based structural damage identification
problem based on a hybrid objective function. Engineering Optimization, 1-19.
FAN, W. & QIAO, P. 2011. Vibration-based damage identification methods: a review and
comparative study. Structural health monitoring, 10, 83-111.
HAO, H. & XIA, Y. 2002. Vibration-based damage detection of structures by genetic
algorithm. Journal of computing in civil engineering, 16, 222-229.
HOSSEINZADEH, A. Z., AMIRI, G. G., RAZZAGHI, S. S., KOO, K. & SUNG, S. 2016. Structural
damage detection using sparse sensors installation by optimization procedure based
on the modal flexibility matrix. Journal of Sound and Vibration, 381, 65-82.

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121
HU, H., WANG, B.-T., LEE, C.-H. & SU, J.-S. 2006. Damage detection of surface cracks in
composite laminates using modal analysis and strain energy method. Composite
structures, 74, 399-405.
HUAJUN, L., SHUQING, W. & HEZHEN, Y. Modal strain energy decomposition method for
damage detection of an offshore structure using modal testing information. Third
Chinese German joint symposium on coastal and ocean engineering. Tainan, 2006. 8-
16.
KANG, F., LI, J.-J. & XU, Q. 2012. Damage detection based on improved particle swarm
optimization using vibration data. Applied Soft Computing, 12, 2329-2335.
KAVEH, A. 2017. Damage detection in skeletal structures based on CSS optimization using
incomplete modal data. Applications of Metaheuristic Optimization Algorithms in
Civil Engineering. Springer.
KAVEH, A. & DADRAS, A. 2018. Structural damage identification using an enhanced thermal
exchange optimization algorithm. Engineering Optimization, 50, 430-451.
KAVEH, A. & ZOLGHADR, A. 2017. Cyclical Parthenogenesis Algorithm for guided modal
strain energy based structural damage detection. Applied Soft Computing, 57, 250-
264.
KHATIR, S., DEKEMELE, K., LOCCUFIER, M., KHATIR, T. & WAHAB, M. A. 2018. Crack
identification method in beam-like structures using changes in experimentally
measured frequencies and Particle Swarm Optimization. Comptes Rendus
Mécanique.
KHATIR, S., BELAIDI, I., KHATIR, T., HAMRANI, A., ZHOU, Y. L., & ABDEL WAHAB, M. 2017.
Multiple damage detection in composite beams using Particle Swarm
Optimization and Genetic Algorithm. Mechanika, 23(4), 514-521.

KHATIR, S., & WAHAB, M. A. 2018. Fast simulations for solving fracture mechanics inverse
problems using POD-RBF XIGA and Jaya algorithm. Engineering Fracture Mechanics.

KIM, J.-T., RYU, Y.-S., CHO, H.-M. & STUBBS, N. 2003. Damage identification in beam-type
structures: frequency-based method vs mode-shape-based method. Engineering
structures, 25, 57-67.

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121
LIFSHITZ, J. M. & ROTEM, A. 1969. Determination of reinforcement unbonding of
composites by a vibration technique. Journal of Composite Materials, 3, 412-423.
LO IACONO, F., NAVARRA, G. & PIRROTTA, A. 2012. A damage identification procedure
based on Hilbert transform: experimental validation. Structural Control and Health
Monitoring, 19, 146-160.
MIRJALILI, S. 2016. Moth-flame optimization algorithm: A novel nature-inspired heuristic
paradigm. Knowledge-Based Systems, 89, 228-249.
NANDA, B., MAITY, D. & MAITI, D. K. 2014. Crack assessment in frame structures using
modal data and unified particle swarm optimization technique. Advances in
Structural Engineering, 17, 747-766.
PERERA, R., FANG, S.-E. & HUERTA, C. 2009. Structural crack detection without updated
baseline model by single and multiobjective optimization. Mechanical Systems and
Signal Processing, 23, 752-768.
RYTTER, ANDERS. 1993. Vibrational based inspection of civil engineering structures. Dept. of
Building Technology and Structural Engineering, Aalborg University.
SAADA, M. M., ARAFA, M. H. & NASSEF, A. O. 2013. Finite element model updating
approach to damage identification in beams using particle swarm optimization.
Engineering optimization, 45, 677-696.
SALAWU, O. 1997. Detection of structural damage through changes in frequency: a review.
Engineering structures, 19, 718-723.
SAMIR, K., BRAHIM, B., CAPOZUCCA, R. & WAHAB, M. A. 2017. Damage detection in CFRP
composite beams based on vibration analysis using proper orthogonal
decomposition method with radial basis function and Cuckoo Search algorithm.
Composite Structures.
SANDESH, S. & SHANKAR, K. 2010. Application of a hybrid of particle swarm and genetic
algorithm for structural damage detection. Inverse Problems in Science and
Engineering; Formerly Inverse Problems in Engineering, 18, 997-1021.
SAZONOV, E. & KLINKHACHORN, P. 2005. Optimal spatial sampling interval for damage
detection by curvature or strain energy mode shapes. Journal of sound and
vibration, 285, 783-801.

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121
SEYEDPOOR, S. 2012. A two stage method for structural damage detection using a modal
strain energy based index and particle swarm optimization. International Journal of
Non-Linear Mechanics, 47, 1-8.
SHI, Z., LAW, S. & ZHANG, L. 1998. Structural damage localization from modal strain energy
change. Journal of sound and vibration, 218, 825-844.
SHI, Z., LAW, S. & ZHANG, L. M. 2000. Structural damage detection from modal strain energy
change. Journal of engineering mechanics, 126, 1216-1223.
SOHN, HOON, CHARLES R FARRAR, FRANCOIS M HEMEZ, DEVIN D SHUNK, DANIEL W
STINEMATES, BRETT R NADLERr, and JERRY J CZARNECKI. 2003. A review of structural
health monitoring literature: 1996–2001. Los Alamos National Laboratory, USA.
TEUGHELS, A. & DE ROECK, G. 2005. Damage detection and parameter identification by
finite element model updating. Revue européenne de génie civil, 9, 109-158.
TIACHACHT, S., BOUAZZOUNI, A., KHATIR, S., WAHAB, M. A., BEHTANI, A., & CAPOZUCCA, R.
2018. Damage assessment in structures using combination of a modified Cornwell
indicator and genetic algorithm. Engineering Structures, 177, 421-430.

YAN, Y., CHENG, L., WU, Z. & YAM, L. 2007. Development in vibration-based structural
damage detection technique. Mechanical systems and signal processing, 21, 2198-
2211.
YANG, C., HOU, X., WANG, L. & ZHANG, X. 2016. Applications of different criteria in
structural damage identification based on natural frequency and static displacement.
Science China Technological Sciences, 59, 1746-1758.
ZARE HOSSEINZADEH, ABDOLLAH BAGHERI, GHOLAMREZA GHODRATI AMIRI, and KI-YOUNG
KOO. 2014. A flexibility-based method via the iterated improved reduction system
and the cuckoo optimization algorithm for damage quantification with limited
sensors. Smart Materials and Structures, 23: 045019.
ZARE HOSSEINZADEH, A., GHODRATI AMIRI, G. & KOO, K.-Y. 2016. Optimization-based
method for structural damage localization and quantification by means of static
displacements computed by flexibility matrix. Engineering Optimization, 48, 543-561.
ZARE HOSSEINZADEH, A., GHODRATI AMIRI, G., & SEYED RAZZAGHI, S. A. (2017). Model-
based identification of damage from sparse sensor measurements using Neumann series
expansion. Inverse Problems in Science and Engineering, 25(2), 239-259.‫‏‬

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121
ZARE HOSSEINZADEH, A., GHODRATI AMIRI, SA SEYED RAZZAGHI, KY KOO, and SEUNG-HUN
SUNG. 2016. Structural damage detection using sparse sensors installation by
optimization procedure based on the modal flexibility matrix.Journal of Sound and
Vibration, 381: 65-82.
ZHU, H., LI, L. & HE, X.-Q. 2011. Damage detection method for shear buildings using the
changes in the first mode shape slopes. Computers & Structures, 89, 733-743.

Table captions

Table 1. The considered damage scenario in the plane steel truss.

Table 2. The considered damage scenario in the plane steel frame.
Table 3. Physical properties of eight-storey shear frame.
Table 4. The considered damage scenario in the eight-storey shear frame.

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121
Table 1. The considered damage scenario in the plane steel truss.

Damage scenario I Damage scenario II Damage scenario III

Element no. Damage (%) Element No. Damage (%) Element No. Damage (%)
10 15 2 15 4 5
8 10 6 10
9 10
13 20

Table 2. The considered damage scenario in the plane steel frame.

Damage scenario I Damage scenario II Damage scenario III
Element no. Damage (%) Element No. Damage (%) Element No. Damage (%)
3 10 2 10 1 20
12 5 5 10 6 15
14 15 11 10
15 25

Table 3. Physical properties of eight-storey shear frame.

Storey Stiffness(KN/m) Mass(kg)
1 6 300
2 4 200
3 3.5 250
4 3 300
5 5 400
6 4 250
7 5.5 400
8 2.5 300

Table 4. The considered damage scenario in the eight-storey shear frame.

Damage scenario I Damage scenario II Damage scenario III
Story no. Damage (%) Story no. Damage (%) Story no. Damage (%)
2 10 5 5 1 10
7 15 3 10
6 20

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121
Figure captions

Figure 1. Finite-element model of plane steel truss (Zare Hosseinzadeh et al. 2016)
Figure 2. Damage detection results for first damage scenario in plane steel truss considering
0% and 5% noise using (a) first mode shape's data, and (b) first three mode shapes'
data.
Figure 3. Damage detection results for second damage scenario in plane steel truss
considering 0% and 5% noise using (a) first mode shape's data, and (b) first three
mode shapes' data.
Figure 4. Damage detection results for third damage scenario in plane steel truss considering
0% and 5% noise using (a) first mode shape's data, and (b) first three mode shape's
data.
Figure 5. Time of run for plane steel truss using one mode's data in 1000 iteration, with 5%
noise
Figure 6. Convergence plot for steel truss in third scenario with a) 0% noise and b) 5% noise,
m denotes the number of using modes
Figure 7. Detection of damage using PSO algorithm for third scenario of plane steel truss
using one mode's data
Figure 8. Detection of damage using CS algorithm for third scenario of plane steel truss
using one mode's data
Figure 9. Comparison of Convergence plot between MFO, PSO, and CS algorithms in plane
steel truss for third scenario using one mode's data with a) 0% noise and b) 5% noise
Figure 10. Time of run comparison for steel plane truss in third scenario using one mode's
data, optimizing by MFO, PSO, and CS algorithms
Figure 11. Finite-element model of plane steel frame
Figure 12. Damage detection results for first damage scenario in plane steel frame
considering 0% and 5% noise using (a) first mode shape's data (b) first three mode
shapes' data.
Figure 13. damage detection results for second damage scenario in plane steel frame
considering 0% and 5% noise using (a) first mode shape's data (b) first three mode
shapes' data.
Figure 14. damage detection results for third damage scenario in plane steel frame
considering 0% and 5% noise using (a) first mode shape's data (b) first three mode
shapes' data.
Figure 15. Time of run for plane steel frame using one mode's data in 1000 iteration, with
5% noise
Figure 16. Convergence plot for plane steel frame in third scenario with a) 0% noise and b)
5% noise, m denotes the number of using modes
Figure 17. Finite-element model of shear frame.
Figure 18. Damage detection results for first damage scenario in shear frame considering
0%, 5% and 7% noise using (a) first mode shape's data, and (b) first three mode
shapes' data.
Figure 19. Damage detection results for second damage scenario in shear frame considering
0%, 5% and 7% noise using (a) first mode shape's data, and (b) first three mode
shapes' data
Figure 20. Damage detection results for third damage scenario in shear frame considering
0%, 5% and 7% noise using (a) first mode shape's data, and (b) first three mode
shapes' data

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121
Figure 21. Time of run for shear frame using one mode's data in 200 iteration, with 7% noise
Figure 22. Convergence plot for shear frame in third scenario with a) 0% noise, b) 5% noise
and c) 7% noise. m denotes the number of using modes.

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.
 Accepted manuscript doi:
10.1680/jstbu.18.00121

Downloaded by [ La Trobe University] on [17/03/19]. Copyright © ICE Publishing, all rights reserved.