TY - GEN
T1 - Identification of time-varying parameters using variational Bayes-sequential ensemble Monte Carlo sampler
AU - Lye, Adolphus
AU - Gray, Ander
AU - Patelli, Edoardo
N1 - Funding Information: The authors would like to express their gratitude to the Singapore Nuclear Research and Safety Initiatives (SNRSI), EPSRC iCase studentship award 15220067, and Euratom research grant under agreement No. 633053, for their continued support in making this research possible.
Publisher Copyright: © ESREL 2021. Published by Research Publishing, Singapore.
PY - 2021/9/19
Y1 - 2021/9/19
N2 - This work presents an extended sequential Monte Carlo sampling algorithm embedded with a Variational Bayes step. The algorithm is applied to estimate the distribution of time-varying parameters in a Bayesian filtering procedure. This algorithm seeks to address the case whereby the state-evolution model does not have an inverse function. In the proposed approach, a Gaussian mixture model is adopted whose covariance matrix is determined via principle component analysis. As a form of verification, a numerical example involving the identification of inter-storey stiffness within a 2-DOF shear building model is presented whereby the stiffness parameters degrade according to a simple State-evolution model whose inverse function can be derived. The Variational Bayes-sequential ensemble Monte Carlo sampler is implemented alongside the Sequential Monte Carlo sampler and the results compared on the basis of the accuracy and precision of the estimates as well computational time. A non-linear time-series model whose state-evolution model does not yield an inverse function is also analysed to show the applicability of the proposed approach.
AB - This work presents an extended sequential Monte Carlo sampling algorithm embedded with a Variational Bayes step. The algorithm is applied to estimate the distribution of time-varying parameters in a Bayesian filtering procedure. This algorithm seeks to address the case whereby the state-evolution model does not have an inverse function. In the proposed approach, a Gaussian mixture model is adopted whose covariance matrix is determined via principle component analysis. As a form of verification, a numerical example involving the identification of inter-storey stiffness within a 2-DOF shear building model is presented whereby the stiffness parameters degrade according to a simple State-evolution model whose inverse function can be derived. The Variational Bayes-sequential ensemble Monte Carlo sampler is implemented alongside the Sequential Monte Carlo sampler and the results compared on the basis of the accuracy and precision of the estimates as well computational time. A non-linear time-series model whose state-evolution model does not yield an inverse function is also analysed to show the applicability of the proposed approach.
KW - Bayesian model updating
KW - Gaussian mixture model
KW - Markov model
KW - sequential Monte Carlo
KW - uncertainty quantification
KW - variational bayes
UR - http://www.scopus.com/inward/record.url?scp=85135479209&partnerID=8YFLogxK
U2 - 10.3850/978-981-18-2016-8_081-cd
DO - 10.3850/978-981-18-2016-8_081-cd
M3 - Conference contribution book
AN - SCOPUS:85135479209
SN - 9789811820168
T3 - Proceedings of the 31st European Safety and Reliability Conference, ESREL 2021
SP - 443
EP - 450
BT - Proceedings of the 31st European Safety and Reliability Conference, ESREL 2021
A2 - Castanier, Bruno
A2 - Cepin, Marko
A2 - Bigaud, David
A2 - Berenguer, Christophe
T2 - 31st European Safety and Reliability Conference, ESREL 2021
Y2 - 19 September 2021 through 23 September 2021
ER -