Nowadays the neural network has been developed very well and fits into any daily application. In this study, a specific class of a neural network, namely Autoencoder would be used to demonstrate its ability in monitoring the condition of a structure. As well as the other neural network, there is no exact rule to dtermine the architect of the Autoencoder. A traditional way to determine this architect is only focusing on regenerate target feature as well as possible. In the structural health monitoring point of view, the objective is not only limited to the regenerate the feature target well, but also to distinguished the healthy and unhealthy condition of a structure with the high accuracy. Therefore, this study purpose the specificity and sensitivity index combine with the Gaussian distribution random number as a tool to determine the architecture of the Autoencoder. The Gaussian distribution random number would represent the damage condition on the structure, while the sensitivity index would represent the success rate of a model in identifying the damage condition of a structure. Through this index, the most suitable model for a certain feature target was selected. The trend of structural features due to varying environmental factors are simulated by both linear and nonlinear mathematical models. Then the elected model by different kind of index was compared here. Besides, the application to long-term monitoring of natural frequency of Sayano-Shushenskaya Dam and deformation of Fei-Tsui Dam are also studied. The finite element model of both dams is constructed and then the change of behavior due to several crack patterns are considered. The results indicate the selected architect model by proposed approach demonstrate a relatively higher performance compare to the selected model by the traditional approach. Moreover, the proposed approach succeed to expose a model with certain number of nodes in the bottleneck layer that did not qualified for structural health monitoring task.

Nowadays the neural network has been developed very well and fits into any daily application. In this study, a specific class of a neural network, namely Autoencoder would be used to demonstrate its ability in monitoring the condition of a structure. As well as the other neural network, there is no exact rule to dtermine the architect of the Autoencoder. A traditional way to determine this architect is only focusing on regenerate target feature as well as possible. In the structural health monitoring point of view, the objective is not only limited to the regenerate the feature target well, but also to distinguished the healthy and unhealthy condition of a structure with the high accuracy. Therefore, this study purpose the specificity and sensitivity index combine with the Gaussian distribution random number as a tool to determine the architecture of the Autoencoder. The Gaussian distribution random number would represent the damage condition on the structure, while the sensitivity index would represent the success rate of a model in identifying the damage condition of a structure. Through this index, the most suitable model for a certain feature target was selected. The trend of structural features due to varying environmental factors are simulated by both linear and nonlinear mathematical models. Then the elected model by different kind of index was compared here. Besides, the application to long-term monitoring of natural frequency of Sayano-Shushenskaya Dam and deformation of Fei-Tsui Dam are also studied. The finite element model of both dams is constructed and then the change of behavior due to several crack patterns are considered. The results indicate the selected architect model by proposed approach demonstrate a relatively higher performance compare to the selected model by the traditional approach. Moreover, the proposed approach succeed to expose a model with certain number of nodes in the bottleneck layer that did not qualified for structural health monitoring task.

REFERENCES
(2010). "Correction to "Diagnostic methods I: sensitivity, specificity, and other measures of accuracy"." Kidney International, 77(4).
Agatonovic-Kustrin, S., and Beresford, R. (2000). "Basic concepts of artificial neural network (ANN) modeling and its application in pharmaceutical research." Journal of Pharmaceutical and Biomedical Analysis, 22(5), 717-727.
Akaike, H. (1969). "Fitting autoregressive models for prediction." Annals of the Institute of Statistical Mathematics, 21(1), 243-247.
Altman, D. G., and Bland, J. M. (1994). "Statistics Notes: Diagnostic tests 1: sensitivity and specificity." BMJ, 308(6943), 1552.
Arangio, S., and Bontempi, F. (2015). "Structural health monitoring of a cable-stayed bridge with Bayesian neural networks." Structure and Infrastructure Engineering, 11(4), 575-587.
Baldi, P. (2012). "Autoencoders, Unsupervised Learning, and Deep Architectures." Proceedings of ICML Workshop on Unsupervised and Transfer Learning, G. Isabelle, D. Gideon, L. Vincent, T. Graham, and S. Daniel, eds., PMLR, Proceedings of Machine Learning Research, 37--49.
Bendat, J. S., and Piersol, A. G. (2010). Random Data.
Bukenya, P., Moyo, P., Beushausen, H., and Oosthuizen, C. (2014). "Health monitoring of concrete dams: a literature review." Journal of Civil Structural Health Monitoring, 4(4), 235-244.
Caicedo, J. M., Dyke, S. J., and Johnson, E. A. (2004). "Natural Excitation Technique and Eigensystem Realization Algorithm for Phase I of the IASC-ASCE Benchmark Problem: Simulated Data." Journal of Engineering Mechanics, 130(1), 49-60.
Chen, L. M. (1988). "Application of an alternative method including foundation effect for arch dam seismic analysis." Proc., The Ninth World Conference on Earthquake Engineering, 349-354.
Cybenko, G. (1989). "Approximation by superpositions of a sigmoidal function." Mathematics of Control, Signals, and Systems, 2(4), 303-314.
Daszykowski, M., Walczak, B., and Massart, D. L. (2003). "A journey into low-dimensional spaces with autoassociative neural networks." Talanta, 59(6), 1095-1105.
Drobatz, K. J. (2009). "Measures of accuracy and performance of diagnostic tests." J Vet Cardiol, 11 Suppl 1, S33-40.
Emanov, A. F., Seleznev, V. S., Bakh, A. A., Gritsenko, S. A., Danilov, I. A., Kuz'menko, A. P., Saburov, V. S., and Tat'kov, G. I. (2002). "Standing waves in engineering seismology." Geol Geofiz, 43(2), 192-207.
Foresee, F. D., and Hagan, M. T. "Gauss-Newton approximation to Bayesian learning." Proc., Neural Networks,1997., International Conference on, 1930-1935 vol.1933.
Giraldo, D. F., Dyke, S. J., and Caicedo, J. M. (2006). "Damage detection accommodating varying environmental conditions." Struct Health Monit, 5(2), 155-172.
Giurgiutiu, V. (2014). Structural Health Monitoring with Piezoelectric Wafer Active Sensors.
Graham, W. J., and Reclamation, U. S. B. o. (1999). A Procedure for Estimating Loss of Life Caused by Dam Failure, U.S. Department of the Interior, Bureau of Reclamation.
Hinton, G. E., Osindero, S., and Teh, Y.-W. (2006). "A fast learning algorithm for deep belief nets." Neural Comput., 18(7), 1527-1554.
Hinton, G. E., and Salakhutdinov, R. R. (2006). "Reducing the dimensionality of data with neural networks." Science, 313(5786), 504-507.
Kramer, M. A. (1991). "Nonlinear principal component analysis using autoassociative neural networks." AIChE Journal, 37(2), 233-243.
Liseikin, A., Seleznev, V. S., and Krechetov, D. (2014). "Об изменении значений собственных частот плотины Саяно-Шушенской ГЭС при различных уровнях наполнения водохранилища." Proc., Всероссийской конференции, посвященной 100-летию со дня рождения академика Н.Н. Пузырева, Изд-во ИНГГ СО РАН, Новосибирск, 182-186.
Loh, C.-H., and Wu, T.-S. (1996). "Identification of Fei-Tsui arch dam from both ambient and seismic response data." Soil Dynamics and Earthquake Engineering, 15(7), 465-483.
MacKay, D. J. C. (1992). "Bayesian Interpolation." Neural Computation, 4(3), 415-447.
Mandrekar, J. N. (2010). "Simple statistical measures for diagnostic accuracy assessment." J Thorac Oncol, 5(6), 763-764.
Manson, G. "Identifying damage sensitive, environment insensitive features for damage detection." Proc., Proceedings of the third international conference on identification in engineering systems, 187-197.
Nielsen, M. A. (2015). "Neural Networks and Deep Learning." Determination Press.
Peeters, B., and De Roeck, G. (2001). "One-year monitoring of the Z24-Bridge: environmental effects versus damage events." Earthquake Eng Struc, 30(2), 149-171.
Pramodh, C., and Ravi, V. (2007). "Modified great deluge algorithm based auto associative neural network for bankruptcy prediction in banks." International journal of computational intelligence research, 3(4), 363-371.
Rich, T. P. (2006). "Lessons in social responsibility from the Austin dam failure." Int J Eng Educ, 22(6), 1287-1296.
Rumelhart, D. E., Hinton, G. E., and Williams, R. J. (1986). "Learning internal representations by error propagation." Parallel distributed processing: explorations in the microstructure of cognition, vol. 1, E. R. David, L. M. James, and C. P. R. Group, eds., MIT Press, 318-362.
Rytter, A. (1993). "Vibrational Based Inspection of Civil Engineering Structures." Ph.D, Aalborg University, Aalborg.
Sang-Dong, J., and Jaehwan, K. (2012). "Passive wireless structural health monitoring sensor made with a flexible planar dipole antenna." Smart Materials and Structures, 21(2), 027001.
Sanger, T. D. (1989). "Optimal unsupervised learning in a single-layer linear feedforward neural network." Neural Networks, 2(6), 459-473.
Sohn, H., Dzwonczyk, M., Straser, E. G., Kiremidjian, A. S., Law, K. H., and Meng, T. (1999). "An experimental study of temperature effect on modal parameters of the Alamosa Canyon Bridge." Earthquake Eng Struc, 28(8), 879-897.
Sohn, H., Park, G., Wait, J. R., Limback, N. P., and Farrar, C. R. (2004). "Wavelet-based active sensing for delamination detection in composite structures." Smart Materials and Structures, 13(1), 153.
Sohn, H., Worden, K., and Farrar, C. R. (2001). "Novelty detection under changing environmental conditions." P Soc Photo-Opt Ins, 4330, 108-118.
Stewart, M. G., Wang, X., and Nguyen, M. N. (2011). "Climate change impact and risks of concrete infrastructure deterioration." Engineering Structures, 33(4), 1326-1337.
Tannant, D. D. (2015). "Changes to dam safety management in British Columbia triggered by failure of a small earth dam." Georisk: Assessment and Management of Risk for Engineered Systems and Geohazards, 9(3), 148-157.
Tannant, D. D., and Skermer, N. (2013). "Mud and debris flows and associated earth dam failures in the Okanagan region of British Columbia." Canadian Geotechnical Journal, 50(8), 820-833.
van Stralen, K. J., Stel, V. S., Reitsma, J. B., Dekker, F. W., Zoccali, C., and Jager, K. J. (2009). "Diagnostic methods I: sensitivity, specificity, and other measures of accuracy." Kidney Int, 75(12), 1257-1263.
Vidakovic, B. (2011). "Sensitivity, Specificity, and Relatives." Statistics for Bioengineering Sciences: With MATLAB and WinBUGS Support, Springer New York, New York, NY, 109-130.
Wiersma, P. (1989). "The response-spectrum dynamic analysis of gravity dams using the finite element method : phase II." US Army Corps of Engineers, Vicksburg, Mississippi.
Yan, A. M., Kerschen, G., De Boe, P., and Golinval, J. C. (2005). "Structural damage diagnosis under varying environmental conditions—Part I: A linear analysis." Mechanical Systems and Signal Processing, 19(4), 847-864.
Yapar, O., Basu, P. K., Volgyesi, P., and Ledeczi, A. (2015). "Structural health monitoring of bridges with piezoelectric AE sensors." Engineering Failure Analysis, 56, 150-169.
Zurada, J. (1992). Introduction to artificial neural systems, West Publishing Co.
Гольдин, С. В. (1974). Линейные преобразования сейсмических сигналов, Недра.