318-311 3 16 1395

mme.modares.ac.ir

6 5 4 3 *2 1

-1
-2
-3
-4
-5

D
( ) -6
( ) -7
ah_banakar@modares.ac.ir 14115- 111 : *

.
.
SI .
1394
1394
07 :
17 :
17 :
of
1395

(ANN ) .
.
. %97 %98 %98
ve

(R2)
(ANN)
(RMSE ) . 0.87 0.91 0.87
. % 5.6 %5.4 %3.6
hi

Analysis of Timing Belt Vibrational Behavior During a Durability Test Using

Artificial Neural Network (ANN)
c

Meghdad Khazaee1, Ahmad Banakar1*, Barat Ghobadian1, Mostafa Mirsalim2, Saeid Minaei1, Seyed
Ar

Mohamad Jafari3, Peyman Sharghi3

1- Department of Biosystems Engineering, Tarbiat Modares University, Tehran, Iran
2- Department of Mechanical Engineering, Amirkabir University of Technology, Tehran, Iran
3- Irankhodro Powertrain Company, Tehran, Iran
* P.O.B. 14115-111, Tehran, Iran, ah_banakar@modares.ac.ir

ARTICLE INFORMATION ABSTRACT

Original Research Paper In this research, an intelligent method is introduced for prediction of remaining useful life of an internal
Received 08 November 2015 combustion engine timing belt based on its vibrational signals. For this goal, an accelerated durability
Accepted 06 February 2016 test for timing belt was designed and performed based on high temperature and high pre tension. Then,
Available Online 26 March 2016
the durability test was began and vibration signals of timing belt were captures using a vibrational
Keywords:
displacement meter laser device. Three feature functions, namely, Energy, Standard deviation and
Timing belt kurtosis were extracted from vibration signals of timing belt in healthy and faulty conditions and timing
Remaining Useful Life (RUL) belt failure threshold was determined. The Artificial Neural Network (ANN) was used for predicting
Accelerated durability test and monitoring vibrational behavior of timing belt. Finally, the ANN based on Energy, Standard
Vibration Signal
deviation and kurtosis features of vibration signals could predict timing belt remaining useful life with
Artificial Neural Networks (ANN)
accuracy of 98%, 98% and 97%, respectively. The correlation factor (R2) of vibration time series
prediction by ANN and based on Energy, Standard deviation and kurtosis features of vibration signals
were determined as 0.87, 0.91 and 87, respectively. Also, Root Mean Square Error (RMSE) of ANN
based on Energy, Standard deviation and kurtosis features of vibration signals was calculated as 3.6%,
5.4% and 5.6%, respectively.

Please cite this article using: :

M. Khazaee, A. Banakar, B. Ghobadian, M. Mirsalim, S. Minaei, S. M. Jafari, P. Sharghi, Analysis of Timing Belt Vibrational Behavior During Durability Test Using Artificial
Neural Network (ANN), Modares Mechanical Engineering Vol. 16, No. 3, pp. 311-318, 2016 (in Persian)

www.SID.ir
 (SVM) -1
.
.
. 3

. 98
.[1]
.[10]
(2012) .
(RUL) .
4
PHM .
. .
. .[2]
.
.
.[3]

D
.[11] .
(2012)

SI
[6] [5] ( ) [4,2]
. [7]
.[12]
(2013)
of
. 1
.

- . .
- . .[3]
ve

.[13] .

. .
hi

. .
c

. .[8]
Ar

-2 - (2009)
. - 2

. -
.
.

. .[9]
3 (2011)
(STD) .

3 1
Prognostics Center of Excellence (PCoE), University of Cincinnati. Timing belt
4 2
Bayesian theory Dempster-Shafer evidence theory

3 16 1395 312

www.SID.ir
 1 5 4 3
. 1 3 .
.
.
4
. 10 .

.
1 4 .[14]
.
. M 1
N x(n) 1 .

.
.
.[15]
.[17,16]

D
(STD)
-4

SI
.
. [19] 2
[18]
[20]
2 .
of
Table 2 Specification of the laser vibrometer sensor
6
[21]
YP06MGV80
20 [22]
ve

5
660
0.5
-3
1
hi

5000
1000 .
500
c
Ar

.
3

1
Table 1 Name and formulas of the extracted features from vibration
signals

= ( ( ))
Fig. 1 The proposed method in this research for timing belt life
prediction ( ( ) )
=
1 1
( ( ) )
=
3 ( 1)
Root Mean Square (RMS)
4
Kurtosis
5 1
Impulse factor Data Point
6 2
Wenglor Plexiglas

313 3 16 1395

www.SID.ir
 .
. .[23]
. 530 345
3 2 .[24]

-2-5 . ( )
- 16 4 .
.[26] 5500 . 1.6
. 110
.
. 120
60 2 .
.[27]
-5

D
(ANN) -6 -1-5

SI
. .[25]
1

. .
of
.[29,28]
2
(MLP)
( )
3
MLP .
ve

- . 4
- .
MLP
.[31,30]
hi

.
.
c
Ar

Fig. 3 Overloading condition in the timing belt

3
Fig. 2 Experimental test rig of this research
2
3
Multi-Layer Perceptron (MLP) 2
Supervised Learning
4
Levenberg-Marquardt
5 1
Targets Tensioner

3 16 1395 314

www.SID.ir
 . 10 1
. ( )
. 11
.
.[30]
2
MLP
-7 .
-1-7 . 4 3
MLP
4 MLP
4 . .
84 MLP
. .

MLP
-2-7 .
180 . 5
LM
2 3 . 2

D
7 5 . .

SI
3 .
MLP .
. .
I . I*H*O MLP
of
4 .
. I
-3-7
. O . 3
ve

10 8 .
.
hi

2
Vibration amplitude (cm)

H .
1
( )
.
c

0
( )
Ar

-1

100 10 .[32]
-2
0 1000
Time (minute)
2000 3000 4000 5000
.
Fig. 4 Run to failure vibration signals of timing belt during durability
test
4 .
6
20 60
3 . 8
20 7

Table 3 Thresholding of timing belt failure based on vibration features 9

of timing belts
.

1481 1424 467 1

Back Propagation
2
Sigmoid
0.37 0.35 0.09 3
Neural Network Toolbox
4
5.31 5.1 2.48 MATLAB
5
Levenberg-Marquardt
6
Train
7
Validation
10 8
Gradient Test
11 9
Batch Mode Overfit

315 3 16 1395

www.SID.ir
 2000 Real data 2000 Healthy timing belt
Predicted data Faulty timing belt
Threshold
1500 1500

Energy
Energy

1000
1000

500
500

0
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Time(minute) 0 20 40 60 80 100 120 140 160 180
Fig. 8 The predicted and real values of energy feature of vibration Number of sample
signals Fig. 5 Thresholding of failure based on energy features of vibration
8 signals
5

0.5 Real data

Predicted data
0.4 Threshold Healthy timing belt
0.5
standard deviation

Faulty timing belt

Standard deviation
0.3 0.4

D
0.2 0.3

0.1 0.2

SI
0.1
0
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Time(minute) 0
Fig. 9 The predicted and real values of standard deviation feature of 0 20 40 60 80 100 120 140 160 180
Number of sample
vibration signals
9 Fig. 6 Thresholding of failure based on standard deviation of vibration
signals
of
6

Real data
6 Predicted data
Threshold 7 Healthy timing belt
Faulty timing belt
5
ve

6
Kurtosis

Kurtosis

5
4
4
3
hi

2
2
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Time(minute) 1
0 20 40 60 80 100 120 140 160 180
c

Fig. 10 The predicted and real values of kurtosis feature of vibration Number of sample
signals Fig. 7 Thresholding of failure based on kurtosis of vibration signals
10
Ar

.
3
10 .[21] 6
10 8
. 2.5 .
5.5
4 . .

4 . .[20]
3
4 . .
.
| |
= 100 × 100 .

3 16 1395 316

www.SID.ir
 . 4

. %97 %98 %98

(R2) .

. 0.87 0.91 0.87

%98 1

. %97 %98 % 5.6 %5.4 %3.6

95 MLP
-9 60 .

( ) .

D
.

SI
- 10 [34 33]
[1] N. Li, R. Zhou, Q. Hu, X. Liu, Mechanical fault diagnosis based on
redundant second generation wavelet packet transform, neighborhood rough
set and support vector machine, Mechanical Systems and Signal Processing,
Vol. 28, pp. 608-621, 2012.
[2] A. Soleimani, K. S. Esmaeilzadeh, experimental fault detection of a ball .
of
bearing using the chaotic behavior features of a vibration signal, Modares
Mechanical Engineering, Vol. 15, No. 2, pp. 289-297, 2015 (in
Persian ).
.
[3] A. K. S. Jardine, D. Lin, D. Banjevic, A review on machinery diagnostics
and prognostics implementing condition-based maintenance, Mechanical
Systems and Signal Processing, Vol. 20, No. 7, pp. 1483-1510, 2006.
.
ve

[4] X. Lou, K. A. Loparo, Bearing fault diagnosis based on wavelet transform

and fuzzy inference, Mechanical Systems and Signal Processing, Vol. 18, No.
5, pp. 1077-1095, 2004. ( )
[5] A. Moosavian, H. Ahmadi, A. Tabatabaeefar, M. Khazaee, Comparison of
two classifiers; K-nearest neighbor and artificial neural network, for fault .
diagnosis on a main engine journal-bearing, Shock and Vibration, Vol. 20,
hi

No. 2, pp. 263-272, 2013.

[6] A. Saghafi, A. Farshidianfar, Bifurcation and chaos control in a gear .[33]
transmission, Modares Mechanical Engineering, Vol. 14, No. 14, pp. 61-68,
2015 (in Persian ). .
[7] K. Mollazade, H. Ahmadi, M. Omid, R. Alimardani, Vibration-based fault
c

diagnosis of hydraulic pump of tractor steering system by using energy

technique, Modern Applied Science, Vol. 3, No. 6, pp. 59-66, 2009.
Ar

[8] J. Z. Sikorska, M. Hodkiewicz, L. Ma, Prognostic modelling options for

remaining useful life estimation by industry, Mechanical Systems and Signal -8
Processing, Vol. 25, No. 5, pp. 1803-1836, 2011.
[9] G. Niu, B.-S. Yang, Dempster–Shafer regression for multi-step-ahead time-
series prediction towards data-driven machinery prognosis, Mechanical
Systems and Signal Processing, Vol. 23, No. 3, pp. 740-751, 2009.
[10] A. Widodo, B.-S. Yang, Machine health prognostics using survival 4
probability and support vector machine, Expert Systems with Applications,
Vol. 38, No. 7, pp. 8430-8437, 2011.
[11] C. Hu, B. D. Youn, P. Wang, J. Taek Yoon, Ensemble of data-driven Table 4 The accuracy of remaining useful life prediction of timing belt
prognostic algorithms for robust prediction of remaining useful life, based on vibration features
Reliability Engineering & System Safety, Vol. 103, pp. 120-135, 2012.
[12] J. Liu, W. Wang, F. Ma, Y. B. Yang, C. S. Yang, A data-model-fusion
prognostic framework for dynamic system state forecasting, Engineering )
Applications of Artificial Intelligence, Vol. 25, No. 4, pp. 814-823, 2012. (%)
[13] B. Chen, P. C. Matthews, P. J. Tavner, Wind turbine pitch faults prognosis
using a-priori knowledge-based ANFIS, Expert Systems with Applications, R2 RMSE
Vol. 40, No. 17, pp. 6863-6876, 2013.
[14] G. Vachtsevanos, F. Lewis, M. Roemer, A. Hess, B. wu, Intelligent fault % 98 4947 4859 0.87 %3.6 44
diagnosis and prognosis for engineering systems, New Jersey: John Wiley &
Sons, pp. 34-49, 2006.
[15] A. Hamilton, M. Fattah, F. Campean, A. Day, Analytical Life Prediction % 98 4809 4756 0.91 %5.4 36
Modelling of an Automotive Timing Belt, SAE SP, Vol. 2184, pp. 137-137,
2008. % 97 4869 4738 0.87 %5.6 33
[16] K. Hashimoto, M. Oyama, N. Watanabe, K. Komatsu, Y. Todani, Highly
Saturated Nitrile Elastomer (HSN) Automotive Applications III, SAE % 97.6 :
Technical Paper, pp. 1988.
1
Root Mean Square Error (RMSE)

317 3 16 1395

www.SID.ir
 belt life laws and a user design guide, Proceedings of the Institution of
Mechanical Engineers, Part D: Journal of Automobile Engineering, Vol. 212,
No. 5, pp. 409-419, 1998.
[26] S. Sundararaman, J. Hu, J. Chen, K. Chandrashekhara, Temperature
dependent fatigue-failure analysis of V-ribbed serpentine belts, International
Journal of Fatigue, Vol. 31, No. 8, pp. 1262-1270, 2009.
[27] R. Perneder, I. Osborne, Handbook Timing Belts: Springer, 2012.
[28] F. Ahmadzadeh, J. Lundberg, Remaining useful life prediction of grinding
mill liners using an artificial neural network, Minerals Engineering, Vol. 53,
pp. 1-8, 2013.
[29] L. Ai, J. Wang, X. Wang, Multi-features fusion diagnosis of tremor based on
artificial neural network and D–S evidence theory, Signal Processing, Vol.
88, No. 12, pp. 2927-2935, 2008.
[30] V. Singh, I. Gupta, H. Gupta, ANN-based estimator for distillation using
Levenberg–Marquardt approach, Engineering Applications of Artificial
Intelligence, Vol. 20, No. 2, pp. 249-259, 2007.
[31] U. Siripatrawan, P. Jantawat, A novel method for shelf life prediction of a
packaged moisture sensitive snack using multilayer perceptron neural
network, Expert Systems with Applications, Vol. 34, No. 2, pp. 1562-1567,
2008.
[32] K. Worden, Structural fault detection using a novelty measure, Journal of
Sound and Vibration, Vol. 201, No. 1, pp. 85-101, 1997.
[33] T. Benkedjouh, K. Medjaher, N. Zerhouni, S. Rechak, Remaining useful life
estimation based on nonlinear feature reduction and support vector
regression, Engineering Applications of Artificial Intelligence, Vol. 26, No. 7,
pp. 1751-1760, 2013.
[34] P. Wang, B. D. Youn, C. Hu, A generic probabilistic framework for
structural health prognostics and uncertainty management, Mechanical
Systems and Signal Processing, Vol. 28, pp. 622-637, 2012.
of
ve
c hi
Ar
3 16 1395
[17] K. Dalgarno, A. Day, R. Moore, Lifetime dependant properties of automotive
power transmission belts, in Proceeding of European seminar on belt drives,
Vol. 172, 1994.
[18] V. T. Tran, B.-S. Yang, M.-S. Oh, A. C. C. Tan, Machine condition
prognosis based on regression trees and one-step-ahead prediction,
Mechanical Systems and Signal Processing, Vol. 22, No. 5, pp. 1179-1193,
2008.
[19] M. S. Hughes, A comparison of Shannon entropy versus signal energy for
acoustic detection of artificially induced defects in Plexiglas, The Journal of
the Acoustical Society of America, Vol. 91, No. 4, pp. 2272-2275, 1992.
[20] K. R. Al-Balushi, B. Samanta, Gear fault diagnosis using energy-based
features of acoustic emission signals, Proceedings of the Institution of
Mechanical Engineers, Part I: Journal of Systems and Control Engineering,
Vol. 216, No. 3, pp. 249-263, 2002.
[21] T. Barszcz, R. B. Randall, Application of spectral kurtosis for detection of a
tooth crack in the planetary gear of a wind turbine, Mechanical Systems and
Signal Processing, Vol. 23, No. 4, pp. 1352-1365, 2009.
[22] C. Pachaud, R. Salvetat, C. Fray, Crest factor and kurtosis contributions to
identify defects inducing periodical impulsive forces, Mechanical Systems
and Signal Processing, Vol. 11, No. 6, pp. 903-916, 1997.
[23] M. Ucar, R. E. Ergun, A. Cengiz, A novel failure diagnosis system design for
automotive timing belts, Experimental Techniques, Vol. 38, No.5, pp. 1-6,
2012.
[24] P. Castellini, E. Cupido, N. Paone, E. Tomasini, Tracking laser doppler
vibrometer for linear motion: application to a timing belt, in Proceeding
of In 4th International Conference on Vibration Measurement by Laser
D
Techniques, pp. 194-200, 2000.
[25] T. H. C. Childs, K. W. Dalgarno, A. J. Day, R. B. Moore, Automotive timing
SI
318
www.SID.ir