Structural health monitoring of a footbridge using Echo State Networks and NARMAX

Wootton, Adam J.; Butcher, John B.; Kyriacou, Theocharis; Day, Charles; Haycock, P.W.

doi:10.1016/j.engappai.2017.05.014

Cited by 17 publications

(8 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Das et al [1] conducted a comparative study of the effectiveness of different vibration-based damage identification methods and proved that time series analysis outperformed other methods in damage identification with the presence of operational or environmental nuisances. Furthermore, they have been successfully incorporated with SVM [29], hidden Markov model (HMM) [30], and ANNs [7,22,23,31] to identify damage. To avoid false diagnoses, an approach combining sensor-clustering-based time-series analysis with the ANN, which was proposed by Kostić and Gül [22], could successfully determine the existence, location, and relative severity of damage for a footbridge finiteelement model under temperature variations.…”

Section: Damage Identification Under Varying Temperature Effectsmentioning

confidence: 99%

Damage Identification of a Steel Frame Based on Integration of Time Series and Neural Network under Varying Temperatures

Huang

Zhao

et al. 2020

Advances in Civil Engineering

View full text Add to dashboard Cite

The effect of varying temperatures is one of the most important challenges of vibration-based damage identification due to its bigger effects on the structural response than the damage itself. This study presents a methodology incorporating the autoregressive (AR) time series model with two-step artificial neural networks (ANNs) to identify damage under temperature variations. AR coefficients, which are extracted by fitting the AR models to acceleration responses, are however sensitive to temperature changes, resulting in false diagnoses. Thus, two-step ANN models with the inputs of difference in AR coefficients are utilized to compensate the detrimental temperature variations. Finite element (FE) models of a steel-braced frame structure, simulating several damage scenarios with different damage locations and severities at fluctuating temperatures, are used to verify the effectiveness and reliability of this approach. Numerical results indicate that the proposed approach could successfully recognize, locate, and quantify damage by using output-only vibration and temperature data regardless of varying temperatures and noise perturbations.

show abstract

Section: Damage Identification Under Varying Temperature Effectsmentioning

confidence: 99%

Damage Identification of a Steel Frame Based on Integration of Time Series and Neural Network under Varying Temperatures

Huang

Zhao

et al. 2020

Advances in Civil Engineering

View full text Add to dashboard Cite

show abstract

“…Both the ESN and liquid state machine (LSM) are machine learning methods for time-series analysis, which are collectively referred to as reservoir computing methods [29]. Just as its name implies, the ESN is composed of an input layer, hidden layer (i.e., reservoir), and an output layer.…”

Section: The Concept Of Esnmentioning

confidence: 99%

“…The wavelet packet energy is employed as a frequency feature. The expression of wavelet packet energy [29] is written in Equation (11).…”

Section: The Concept Of Esnmentioning

confidence: 99%

Transmission Condition Monitoring of 3D Printers Based on the Echo State Network

Zhang

Cabrera

et al. 2019

Applied Sciences

View full text Add to dashboard Cite

Three-dimensional printing quality is critically affected by the transmission condition of 3D printers. A low-cost technique based on the echo state network (ESN) is proposed for transmission condition monitoring of 3D printers. A low-cost attitude sensor installed on a 3D printer was first employed to collect transmission condition monitoring data. To solve the high-dimensional problem of attitude data, feature extraction approaches were subsequently performed. Based on the extracted features, the ESN was finally employed to monitor transmission faults of the 3D printer. Experimental results showed that the fault recognition accuracy of the 3D printer was obtained at 97.17% using the proposed approach. In addition, support vector machine (SVM), locality preserving projection support vector machine (LPPSVM), and principal component analysis support vector machine (PCASVM) were also used for comparison. The contrast results showed that the recognition accuracies of our method were higher and more stable than that of SVM, LPPSVM, and PCASVM when collecting raw data via the low-cost attitude sensor.

show abstract

“…In recent years many data‐driven methodologies are proposed in the SHM literature with different combinations of features and classifiers 6–10 Usage of support vector machine (SVM) for SHM is becoming popular which is evident from some of its recent applications 11–13 . The reasons for its popularity are some of its advantages which are as follows: (a) The flexibility in choosing the appropriate kernels makes it applicable to a wide variety of datasets.…”

Section: Introductionmentioning

confidence: 99%

On the use of acquisition function‐based Bayesian optimization method to efficiently tune SVM hyperparameters for structural damage detection

Agrawal

Chakraborty

2021

Struct Control Health Monit

View full text Add to dashboard Cite

Summary This article presents a novel application of an acquisition function‐based Bayesian optimization (BO) method to optimally tune the hyperparameters of soft margin support vector machine (SVM) for structural damage detection. The performance of SVM classifiers is known to depend heavily on the selection of hyperparameters. The optimal hyperparameters yield the minimum cross‐validation error. However, no explicit functional relationship exists between the hyperparameters and the cross‐validation error which can be used as an objective function. This BO approach uses the Bayes rule along with the concept of Gaussian process regression to estimate the objective function and an acquisition function to efficiently search the hyperparameter space. The performance of this methodology has been validated by training an optimally tuned SVM to carry out damage detection on an experimental benchmark structure using four different datasets. It was observed that the number of function evaluations to calculate the optimal values of hyperparameters turned out to be below 25 for each of the datasets, which is typically the number of function evaluations required (i.e., the swarm size) in each iteration of particle swarm optimization (PSO). A comparison of BO with PSO in terms of the number of function evaluations revealed its efficiency for the benchmark problem. This result signifies the practical importance of BO in speeding up the process of hyperparameter selection. Finally, some interesting observations regarding the obtained values of box constraint, support‐vector‐location, and the relationship between the convergence curve of BO and learning curves of SVM are discussed.

show abstract

Structural health monitoring of a footbridge using Echo State Networks and NARMAX

Cited by 17 publications

References 24 publications

Damage Identification of a Steel Frame Based on Integration of Time Series and Neural Network under Varying Temperatures

Damage Identification of a Steel Frame Based on Integration of Time Series and Neural Network under Varying Temperatures

Transmission Condition Monitoring of 3D Printers Based on the Echo State Network

On the use of acquisition function‐based Bayesian optimization method to efficiently tune SVM hyperparameters for structural damage detection

Contact Info

Product

Resources

About