doi:10.1016/j.engstruct.2007.01.013 
Copyright © 2007 Elsevier Ltd All rights reserved.
Damage detection using artificial neural network with consideration of uncertainties
Norhisham Bakhary
, a, 
, Hong Haoa and Andrew J. Deeksa
aSchool of Civil and Resource Engineering, The University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australia
Received 27 October 2006;
revised 8 January 2007;
accepted 8 January 2007.
Available online 12 March 2007.
References and further reading may be available for this article. To view references and further reading you must
purchase this article.
Abstract
Artificial Neural Networks (ANN) have received increasing attention for use in detecting damage in structures based on vibration modal parameters. However, uncertainties existing in the finite element model used and the measured vibration data may lead to false or unreliable output result from such networks. In this study, a statistical approach is proposed to take into account the effect of uncertainties in developing an ANN model. By applying Rosenblueth’s point estimate method verified by Monte Carlo simulation, the statistics of the stiffness parameters are estimated. The probability of damage existence (PDE) is then calculated based on the probability density function of the existence of undamaged and damaged states. The developed approach is applied to detect simulated damage in a numerical steel portal frame model and also in a laboratory tested concrete slab. The effects of using different severity levels and noise levels on the damage detection results are discussed.
Keywords: Damage detection; Neural networks; Uncertainties; Rosenblueth’s point estimate; Random noise; Modal data
Fig. 1. Probability density functions for αj and 
and probability of damage existence, 
.
Fig. 2. Finite element model of the frame.
Fig. 3. First three mode shapes for undamaged, scenario 1 and scenario 2.
Fig. 4. ANN prediction for scenario 1 and scenario 2 compared to the actual value using noise-free input data.
Fig. 5. ANN prediction for scenario 1 and scenario 2 compared to the actual value using noisy input data.
Fig. 6. Monte Carlo simulation result.
Fig. 7. K–S test result for segment 3.
Fig. 8. Mean values and coefficients of variation (COV) of E values in undamaged state.
(a) Sensor location.
(b) Experimental setup.
Fig. 9. Experimental setup and sensor location.
Fig. 10. Segment on the slab.
(a) Crack pattern at level 1.
(b) Crack pattern at level 2.
(c) Crack pattern at level 3.
Fig. 11. Crack patterns for different damage level.
Fig. 12. ANN prediction without considering uncertainties.
Fig. 13. Probability of damages existence for slab using statistical ANN.
Table 1.
Training functions and testing variables used in point estimation method
Table 2.
E values for scenario 1 and scenario 2
Table 3.
First three frequencies values for scenario 1 and scenario 2
Table 4.
Probability of damage existence (%) for every segment of scenario 1 and scenario 2
Table 5.
Probability of damage existence (%) for different damage severities
Table 6.
Probability of damage existence (%) for different combination of uncertainties in training and testing data
Table 7.
Frequencies of the slab at different load level
Abstract
Purchase PDF (1150 K)
E-mail Article
Add to my Quick Links
Cited By in Scopus (1)
Purchase PDF (565 K)
