Abstract
Recent developments tools and techniques for structural health monitoring allow the design of early warning systems for the damage diagnosis and structural assessment. Most methods to damage detection involve vibration data analysis by using identification systems that generally require a mathematical model and much information about the system, such as parameters and states that are mostly unknown. In this paper, a novel frequency domain convolutional neural network (FDCNN) proposed aims to design an identification system for damage detection based on Bouc–Wen hysteretic model. FDCNN, unlike other works, only requires acceleration measurements for damage diagnosis that are very sensitive to environmental noise. In contrast to neural network (NN) and time domain convolutional neural network, FDCNN reduces the computational time required for the learning stage and adds robustness against noise in data. The FDCNN includes random filters in the frequency domain to avoid measurement noise using a spectral pooling operation, which is useful when the system bandwidth is unknown. Incorrect filtering can produce unwanted results, as a shifted and attenuation signal relative to the original. Moreover, FDCNN allows overcoming the parameterization problem in nonlinear systems, which is often difficult to achieve. In order to validate the proposed methodology, a comparison between two different architectures of convolutional neural networks is made, showing that proposed CNN in frequency domain brings better performance in the identification system for damage diagnosis in building structures. Experimental results from reducing scale two-storey building confirm the effectiveness of the proposed.
Similar content being viewed by others
Notes
The energy of the building is estimated using the CNN output together with the velocity of each floor.
References
Abdeljaber O, Avci O, Kiranyaz S, Gabbouj M, Inman DJ (2017) Real-time vibration-based structural damage detection using one-dimensional convolutional neural networks. J Sound Vib 388:154–170
Appana DK, Prosvirin A, Kim J-M (2018) Reliable fault diagnosis of bearings with varying rotational speeds using envelope spectrum and convolution neural networks. Soft Comput 22(20):6719–6729
Arqub OA, Abo-Hammour Z (2014) Numerical solution of systems of second-order boundary value problems using continuous genetic algorithm. Inf Sci 279:396–415
Arqub OA, Mohammed AL-S, Momani S, Hayat T (2016) Numerical solutions of fuzzy differential equations using reproducing kernel Hilbert space method. Soft Comput 20(8):3283–3302
Arqub OA, Al-Smadi M, Momani S, Hayat T (2017) Application of reproducing kernel algorithm for solving second-order, two-point fuzzy boundary value problems. Soft Comput 21(23):7191–7206
Atha DJ, Jahanshahi MR (2018) Evaluation of deep learning approaches based on convolutional neural networks for corrosion detection. Struct Health Monit 17(5):1110–1128
Bouti A, Mahraz MA, Riffi J, Tairi H (2018) A robust system for road sign detection and classification using lenet architecture based on convolutional neural network. Soft Comput 24:6721–6733
Bursi OS, Ceravolo R, Erlicher S, Zanotti Fragonara L (2013) Identification of the hysteretic behaviour of a partial-strength steel-concrete moment-resisting frame structure subject to pseudodynamic tests. Earthq Eng Struct Dyn 41(14):1883–1903
Carden EP, Fanning P (2004) Vibration based condition monitoring: a review. Struct Health Monit 3:355–377
Ceravolo R, Erlicher S, Fragonara LZ (2013) Comparison of restoring force models for the identification of structures with hysteresis and degradation. J Sound Vib 332(26):6982–6999
Cha Y-J, Choi W, Büyüköztürk O (2017) Deep learning-based crack damage detection using convolutional neural networks. Comput Aided Civ Infrastruct Eng 32(5):361–378
Charles RF, Keith W, Michael DT, Gyuhae P, Jonathon N, Douglas EA, Matthew TB, Kevin F (2007) Nonlinear system identification for damage detection. In: Report LA-14353, Los Alamos National Laboratory (LANL), Los Alamos, NM, pp 1–161
Chatzi EN, Smyth AW, Masri SF (2010) Experimental application of on-line parametric identification for nonlinear hysteretic systems with model uncertainty. Struct Saf 32(5):326–337
Chen S, Billings S, Grant P (1990) Non-linear system identification using neural networks. Int J Control 51(6):1191–1214
Chopra AK (1995) Dynamics of structures: theory and applications to earthquake engineering, 1st edn. Prentice-Hall International series
Cooley JW, Lewis PA, Welch PD (1969) The fast Fourier transform and its applications. IEEE Trans Educ 12(1):27–34
Das S, Saha P, Patro S (2016) Vibration-based damage detection techniques used for health monitoring of structures: a review. J Civ Struct Health Monit 6(3):477–507
Doebling SW, Farrar C, Prime MB (1998) A summary review of vibration-based damage identification methods. Shock Vib Dig 30(2):1–34
Eslami E, Choi Y, Lops Y, Sayeed A (2019) A real-time hourly ozone prediction system using deep convolutional neural network. Neural Comput Appl. https://doi.org/10.1007/s00521-019-04282-x
Fan W, Qiao P (2011) Vibration-based damage identification methods: a review and comparative tudy. Struct Health Monit 10:83–111
Farrar C, Doebling S, Nix D (2001) Vibration-based structural damage identification. Philos Trans R Soc 359(1778):131–149
Glorot X, Bengio Y (2010) Understanding the difficulty of training deep feedforward neural networks. In: Proceedings of the thirteenth international conference on artificial intelligence and statistics, pp 249–256
Hibberler RC (2011) Mechanics of materials, 8th edn. Prentice Hall, pp 1–888
Ikhouane FA, MañOsa VC, Rodellar J (2005) Adaptive control of a hysteretic structural system. Automatica 41(2):225–231
Kim Y (2014) Convolutional neural networks for sentence classification. arXiv:1408.5882
Kong X, Cai C-S, Hu J (2017) The state-of-the-art on framework of vibration-based structural damage identification for decision making. Appl Sci 7(5):497–510
Krizhevsky A, Sutskever I, Hinton GE (2012) Imagenet classification with deep convolutional neural networks. In: Advances in neural information processing systems, pp 1097–1105
Lawrence S, Giles CL, Tsoi AC, Back AD (1997) Face recognition: a convolutional neural-network approach. IEEE Trans Neural Netw 8(1):98–113
LeCun Y, Boser B, Denker JS, Henderson D, Howard RE, Hubbard W, Jackel LD (1989) Backpropagation applied to handwritten zip code recognition. Neural Comput 1(4):541–551
Lin Y-Z, Nie Z-H, Ma H-W (2017) Structural damage detection with automatic feature-extraction through deep learning. Comput Aided Civ Infrastruct Eng 32(12):1025–1046
Liu R, Yang B, Zio E, Chen X (2018a) Artificial intelligence for fault diagnosis of rotating machinery: a review. Mech Syst Signal Process 108:32–47
Liu Y, Huang H, Cao J, Huang T (2018b) Convolutional neural networks-based intelligent recognition of chinese license plates. Soft Comput 22(7):2403–2419
Loh C-H, Mao C-H, Huang J-R, Pan T-C (2011) System identification and damage evaluation of degrading hysteresis of reinforced concrete frames. Earthq Eng Struct Dyn 40(6):623–640
Ma F, Zhang H, Bockstedte A, Foliente GC, Paevere P (2004b) Parameter analysis of the differential model of hysteresis. J Appl Mech 71(3):342–349
Ma F, Ng CH, Ajavakom N (2006) On system identification and response prediction of degrading structures. Struct Control Health Monit 13:347–364
Ma S, Cai W, Liu W, Shang Z, Liu G (2019) A lighted deep convolutional neural network based fault diagnosis of rotating machinery. Sensor 19(10):2381
Maia NMM, Silva JMM, Almas EAM, Sampaio RPC (2003) Damage detection in structures: from mode shape to frequency response function methods. Mech Syst Signal Process 17(3):489–498
Modarres C, Astorga N, Droguett EL, Meruane V (2018) Convolutional neural networks for automated damage recognition and damage type identification. Struct Control Health Monit 25:e2230
Pau A, Vestroni F (2013) Vibration assessment and structural monitoring of the basilica of maxentius in rome. Mech Syst Signal Process 41:454–466
Rahai A, Bakhtiari-Nejad F, Esfandiari A (2007) Damage assessment of structure using incomplete measured mode shapes. Struct Control Health Monit 14:808–829
Rippel O, Snoek J, Adams RP (2015) Spectral representations for convolutional neural networks. In: Advances in neural information processing systems, pp 2449–2457
Roux P, Guéguen P, Baillet L, Hamze A (2014) Structural-change localization and monitoring through a perturbation-based inverse problem. Acoust Soc Am 136:2586–2597
Rucevskis S, Janeliukstis R, Akishin P, Chate A (2016) Mode shape-based damage detection in plate structure without baseline data. Struct Control Health Monit 23:1180–1193
Shan J, Shi W, Lu X (2016a) Model-reference health monitoring of hysteretic building structure using acceleration measurement with test validation. Comput Aided Civ Infrastruct Eng 31:449–464
Shin M, Lee J-H (2016) Cnn based lithography hotspot detection. Int J Fuzzy Log Intell Syst 16(3):208–215
Simard PY, Steinkraus D, Platt JC, et al (2003) Best practices for convolutional neural networks applied to visual document analysis. In: ICDAR, vol 3
Sohn H, Farrar C, Hemez F, Devin DS, Daniel WS, Brett RN, Jerry JC (2003) A review of structural health monitoring literature: 1996–2001. In: Los Alamos National Laboratory report, LA-13976-MS, pp 1–331
Udmale SS, Patil SS, Phalle VM, Singh SK (2019) A bearing vibration data analysis based on spectral kurtosis and convnet. Soft Comput 23(19):9341–9359
Vidal F, Navarro M, Aranda C, Enomoto T (2014) Changes in dynamic characteristics of lorca rc buildings from pre- and post-earthquake ambient vibration data. Bull Earthq Eng 12:2095–2110
Vu T-D, Ho N-H, Yang H-J, Kim J, Song H-C (2018) Non-white matter tissue extraction and deep convolutional neural network for alzheimer’s disease detection. Soft Comput 22(20):6825–6833
Wan Z, Wang T, Li S, Zhang Z (2018) A modified particle filter for parameter identification with unknown inputs. Struct Control Health Monit 25:e2268
Wen YK (1976) Method for random vibration of hysteretic system. J Eng Mech Div 102(2):249–263
Yıldırım Ö, Baloglu UB, Acharya UR (2018) A deep convolutional neural network model for automated identification of abnormal eeg signals. Neural Comput Appl. https://doi.org/10.1007/s00521-018-3889-z
Zhao R, Yan R, Chen Z, Mao K, Wang P, Gao RX (2019) Deep learning and its applications to machine health monitoring. Mech Syst Signal Process 115:213–237
Zhu H, Li L, He X-Q (2011) Damage detection method for shear buildings using the changes in the first mode shape slopes. Comput Struct 89(9–10):733–743
Zou Y, Tong L, Steven GP (2000) Vibration-based model dependent damage (delamination) identification and health monitoring for composite structures: a review. J Sound Vib 230:357–378
Acknowledgements
The authors express their thanks to unknown referees for the careful reading and helpful comments. Authors also appreciate the support of Mr. Jesús Meza for their assistance to complete the experiments. This work was supported in part by the project SEP-CINVESTAV No.62. The second author is also grateful for the financial support of CONACYT. Jesús Morales-Valdez acknowledges the support of Programa Catedras-CONACYT. All authors are grateful to CINVESTAV-IPN for the support in this project.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that there is no conflict of interest in this paper.
Ethical approval
This article does not contain any studies with human participants or animals performed by any of the authors.
Additional information
Communicated by V. Loia.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendices
Appendix A: Discrete Fourier transform
Discrete Fourier transform (DFT), denoted by \(\mathcal {F}(\cdot )\), is a powerful tool to convert spatial samples into a sequence of complex-valued samples in the frequency domain. Some important properties of DFT are as follows: It is linear and unitary (Cooley et al. 1969), and its inverse transform is given by \(\mathcal {F}^{-1}(\cdot )=\mathcal {F}(\cdot )^{*}\) which is the conjugate of the transform itself. This last property is useful during the training stage of CNN. A DFT of n-points is defined as \(A=\mathcal {F}(a)\), where \(\mathcal {F}\) can be expressed as a matrix F; this matrix is called a DFT matrix and it is constructed as follows:
where \(\omega =\exp ^{\frac{-2\pi \mathrm {i}}{n}}\).
Slight modification is made in order to ensure the DC frequency component in the center row of the matrix.
Remark 7
In frequency analysis, the convolution operation becomes an element-wise product which makes the analysis easier and direct. The convolution operation between \(a,b\in \mathfrak {R}^{n}\), using the DFT is:
where \(*\) denotes the convolution operation and \(\odot \) is an element-wise product. This product reduces the number of operations compared to the convolution stage in TDCNN and it make the training process even faster.
Appendix B: Time domain CNN for modeling time series
Consider an unknown discrete-time nonlinear system
where y(q) is the scalar output, x(q) the internal state, u(q) the input, \(f(\cdot )\) and \(g(\cdot )\) smooth functions, \(f,g\in C^{\infty }\) .
A nonlinear autoregressive exogenous (NARX) model for (70) is defined as
The system dynamics are represented by the unknown nonlinear difference equation \(\varPhi \), where
y(q) and u(q) within this equation represent measurable output and input for the system, with \(n_{y}\) and \(n_{u}\) the regression order, respectively, which are unknown.
The nonlinear system identification of (71) based on time domain convolutional neural networks (TDCNN) is shown in (73), where \({\hat{y}}_{T}(q)\) is estimation of the real output generated by TDCNN, which is a scalar element.
This is fully connected layer with W as synaptic weights vector and \(\vartheta \) the stacked output of the last subsample layer of TDCNN.
Two more types of layer are introduced in TDCNN. The first layer in TDCNN is a convolutional one, where two operations are made: convolution and an activation function. The convolution operation is
\(\ell \) represent the actual layer, h-filters per layer are used, and each filter is \(K_{h}^{(\ell )}\in R^{f_{\ell }}\). For each element i of \(\chi _{h}^{(\ell )}\), the previous operation is equivalent to
The result of this operation \(\chi _{h}^{(\ell )}\) is called the feature map, which contains features properties of the input, and each filter obtains a different feature. These feature maps go through an activation function, different activation functions are used in neural networks for specific tasks and unique properties (Glorot and Bengio 2010), but the one used is in this papers is the rectified linear unit (ReLU). The output of a convolutional layer is defined by (76)
for the first layer of the CNN, \(y_{h}^{(\ell -1)}\) is the input vector
where \(r_{1}\) and \(r_{2}\) denote the regression order. \(r_{1}\ne n_{y} \) and \(r_{2}\ne n_{u}\).
After a convolutional layer, a subsample layer is followed; this layer is pretended to be used as data reduction stage, so the strongest response from the filters keeps going through the TDCNN.
In the subsample layers, the operation used is the max-pool, which is defined as
The input divided in groups of dimension \(s_{\ell }\) and from each group the highest values remain. The Shrink depends on the layer where it is applied.
Convolutional and subsample layers can be repeated as many times as the application require in the TDCNN. As mentioned earlier, after the last subsample layer, the outputs of each feature map are stacked to create the vector \(\vartheta \)
This helps to manage the last layer in terms of vector and matrices. The complete architecture is shown in Fig. 18.
1.1 Training of time domain CNN using backpropagation
The training of the TDCNN’s parameters is realized by the backpropagation algorithm (BPA), which is used to calculated the gradient of the cost function respect each parameter of the TDCNN, propagating it backward through the network to update these parameters. The cost function is used as a measurement of the performance, and the most frequently cost function for identification is the squared error which measures the difference between the real output and the estimated one.
\(e_{T}(q)\) is the identification error between the TDCNN output and the real output in each instant, i.e., \(e_{T}(q)={\hat{y}}_{T}(q)-y(q)\).
The BPA uses the gradient of the cost function with respect to each parameter in the neural network. To calculate the gradient, it uses the chain rule and then each parameter is updated by the delta rule. In the output layer, the weights are updated as follows:
where \(w_{i}^{(\ell )}\) are the elements of the vector \(W^{(\ell )}\), \(\eta _{T}\) the learning rate defining one for each layer and
\(\vartheta _{i}\) are the elements of vector \(\vartheta \) corresponding to the weight \(w_{i}^{(\ell )}\). To previous layer, the gradient, using chain rule, is
For the subsample layer, an reverse operation of maxpool is used to calculate the gradient
where \(up(\cdot )\) is an operation to increase length of the gradient to match the previous layer and only passing to the positions where the highest response occurs in the forward stage, leaving everything else in zeros. For convolutional layer, the gradient of the cost function with respect to the filters is calculated as
with \(*\) being the convolution operator and
with \(f^{^{\prime }}(\cdot )\) being the derivative of the ReLU operation, that is defined as,
In order to update the filters, delta rule is used, therefore
Finally, to backpropagate the gradient to previous layer of a convolutional layer, the equation is
The operator \(rot180(\cdot )\) is equivalent to use its parameter from bottom to top, just like a flip over.
Appendix C: Multilayer neural network for system modeling
For comparison, a multilayer perceptron (NN for simplicity) is created. This NN consists of one hidden layer with 35 units, which are paired with activation function \(\tanh (\cdot )\); its architecture is shown in Fig. 19.
Consider the nonlinear system to be identify defined in 71 and regard the same input from the CNN described in 35; the output of the units in the hidden layer is defined as:
\(V_{NN}\) are the synaptic weights in the hidden layers written in matrix form, \(X_{NN}\) is the vector output of hidden layer, and each element corresponds to each one of the units in this layer. The output of the NN is
where \(W_{NN}\) are the synaptic weights in the output layer, dimensions match, so the output is scalar. The training of this NN is done with the backpropagation algorithm. For this matter, the cost function to be minimized is defined as
where \(e(q)=\left( {\hat{y}}_{NN}(q)-y(q)\right) ^2 \) and the update law for the synaptic weights in output and hidden layer is defined with the delta rule, i.e.,
and
where \(\eta _{NN}\) is the learning rate for this NN.
Rights and permissions
About this article
Cite this article
Lopez-Pacheco, M., Morales-Valdez, J. & Yu, W. Frequency domain CNN and dissipated energy approach for damage detection in building structures. Soft Comput 24, 15821–15840 (2020). https://doi.org/10.1007/s00500-020-04912-w
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00500-020-04912-w