A new Kalman filter approach for structural parameter tracking: Application to the monitoring of damaging structures tested on shaking-tables

Diaz, Matthieu; Charbonnel, Pierre-Étienne; Chamoin, Ludovic

doi:10.1016/j.ymssp.2022.109529

Cited by 20 publications

(19 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A more efficient automatic tuning of weighting between losses, or extending previous works on the coupling between reduced order modeling techniques and mCRE, 63 could be a research direction to alleviate this concern. Eventually, to efficiently update the model and predict the quantities of interest, the present work could be coupled with previous works on mCRE to perform real‐time online SHM with Kalman filtering 49,64 and it could integrate ideas from Reference 65 in which some layers are frozen during the online training.…”

Section: Discussionmentioning

confidence: 99%

NN‐mCRE: A modified constitutive relation error framework for unsupervised learning of nonlinear state laws with physics‐augmented neural networks

Benady,

Baranger,

Chamoin

2024

Numerical Meth Engineering

View full text Add to dashboard Cite

This article proposes a new approach to train physics‐augmented neural networks with observable data to represent mechanical constitutive laws. To train the neural network and learn thermodynamics potentials, the proposed method does not rely on strain‐stress or strain‐free energy pairs but needs only partial strain or displacement measurements inside the structure. The neural network is trained thanks to an unsupervised procedure in which the modified constitutive relation error (mCRE) is minimized. The mCRE functional provides a bias‐aware data assimilation framework with a rich physical sense as the constitutive relation error (CRE) part can be interpreted as a modeling error continuously defined over the structure, and can be used as a prediction quality in the inference phase. This article also extends previous works on the mCRE by introducing a new minimization procedure in the case of nonlinear state laws. As typical structural health monitoring applications may require that the neural networks should be trained online, an important focus is thus made on automatic and adaptive tuning of sensitive hyperparameters (learning rate, weighting between losses, number of epochs and initialization). It is shown that when the training database is rich enough with respect to the loading cases, the proposed method achieves remarkable performance regarding the quality of the learned model, noise robustness, and low sensitivity to user‐defined hyperparameters. The method is evaluated on two test cases: a non‐quadratic potential in the small strain regime with synthetic optic fiber measurements, and a Mooney–Rivlin model in the hyperelastic case with synthetic digital image correlation observations.

show abstract

Section: Discussionmentioning

confidence: 99%

NN‐mCRE: A modified constitutive relation error framework for unsupervised learning of nonlinear state laws with physics‐augmented neural networks

Benady,

Baranger,

Chamoin

2024

Numerical Meth Engineering

View full text Add to dashboard Cite

show abstract

“…We aim to present the benefits of mCRE-based OSP for mCRE-based model updating in a case study whose geometry and loading conditions are representative of earthquake engineering problems. Inspired from the SMART2013 test campaign that has been recently used for mCRE applications [47,75], we consider the two-story frame structure of Fig. 1 submitted to a tridimensional low-magnitude random ground acceleration.…”

Section: Description Of the Problemmentioning

confidence: 99%

“…Consequently, the proposed mCRE-based OSP appears as an interesting additional asset for the construction of a mCRE-unified framework for SHM or structural dynamics applications (see Fig. 9 for an example on how OSP could be combined to the recent publications of the authors [47,75]). However, there is no doubt that this tool still lacks of maturity to be properly exploited, and further research should address the strong influence of the confidence into measurements coefficient [80].…”

Section: Conclusion and Prospectsmentioning

confidence: 99%

“…Pareto front algorithms may be a first tool for this purpose. Finally, one of the current on-going investigations of the authors concerns the use of mCRE-based OSP for active sensing purposes in order to improve damage detection in cases where the state of the structure is tracked online via data assimilation techniques [75]. Iterative strategies could then be employed to refine the sensor configuration only where needed, i.e.…”

Section: Conclusion and Prospectsmentioning

confidence: 99%

See 1 more Smart Citation

Merging experimental design and structural identification around the concept of modified Constitutive Relation Error in low-frequency dynamics for enhanced structural monitoring

Diaz

Charbonnel

Chamoin

2023

Mechanical Systems and Signal Processing

Self Cite

View full text Add to dashboard Cite

“…Kalman filters are shown to be sensitive to noisy or corrupted measurements [8,9]. Thereby, integrating the mCRE within this stochastic approach provides a cheap, robust and sequential model updating tool named Modified Dual Kalman Filter (MDKF) [10,11].…”

mentioning

confidence: 99%

Model updating with a Modified Dual Kalman Filter acting on distributed strain measurements

Farahbakhsh¹,

Chamoin²,

Poncelet³

2023

XI International Conference on Adaptive Modeling and Simulation

View full text Add to dashboard Cite

Following the advances in measurement technology and its vast availability, mechanical systems and structures are increasingly equipped with sensors to obtain continuous information regarding the system state. Coupled with robust numerical models, this information can be used to build a numerical twin of the structure that is linked to its physical twin via a feedback loop. This results in the concept of Dynamic Data Driven Application Systems (DDDAS) that can predict and control the evolution of the physical phenomena at stake on the structure, as well as dynamically updating the numerical model with the help of real-time measurements [1,2].

show abstract

A new Kalman filter approach for structural parameter tracking: Application to the monitoring of damaging structures tested on shaking-tables

Cited by 20 publications

References 49 publications

NN‐mCRE: A modified constitutive relation error framework for unsupervised learning of nonlinear state laws with physics‐augmented neural networks

NN‐mCRE: A modified constitutive relation error framework for unsupervised learning of nonlinear state laws with physics‐augmented neural networks

Merging experimental design and structural identification around the concept of modified Constitutive Relation Error in low-frequency dynamics for enhanced structural monitoring

Model updating with a Modified Dual Kalman Filter acting on distributed strain measurements

Contact Info

Product

Resources

About