Details
| Original language | English |
|---|---|
| Title of host publication | Proceedings of the 32nd European Safety and Reliability Conference, ESREL 2022 - Understanding and Managing Risk and Reliability for a Sustainable Future |
| Editors | Maria Chiara Leva, Edoardo Patelli, Luca Podofillini, Simon Wilson |
| Pages | 1852-1858 |
| Number of pages | 7 |
| Publication status | Published - 28 Aug 2022 |
| Event | 32nd European Safety and Reliability Conference (ESREL 2022) - Dublin, Ireland Duration: 28 Aug 2022 → 1 Sept 2022 Conference number: 32 |
Abstract
In this paper, the challenge of quantifying the uncertainty in the estimation of power spectral density (stationary and nonstationary) of ground motion processes subject to missing data is addressed. Specifically, to exploit additional information besides the incomplete recording, simulated ground motions are generated by a stochastic finitefault model, with its region-specific parameters (source, attenuation, and site parameters) modeled as probability distributions. Then a Bayesian neural network is constructed to probabilistically learn the temporal patterns from such uncertain time-series data. Epistemic uncertainties on the model parameters of the Bayesian neural network model are learnt via variational inference. Thanks to the probabilistic merit of the Bayesian neural network, an ensemble of reconstructed realizations can be obtained, which leads to a probabilistic power spectrum, with each frequency component represented by a probability distribution. This framework is of great importance to researches such as stochastic structural dynamics, where accurate stochastic representations are needed for characterizing engineering excitation processes but faced with incomplete ground motion recordings.
Keywords
- evolutionary spectral density, ground motion, stochastic processes, uncertainty quantification, variational inference
ASJC Scopus subject areas
- Engineering(all)
- Safety, Risk, Reliability and Quality
- Social Sciences(all)
- Safety Research
Cite this
- Standard
- Harvard
- Apa
- Vancouver
- BibTeX
- RIS
Proceedings of the 32nd European Safety and Reliability Conference, ESREL 2022 - Understanding and Managing Risk and Reliability for a Sustainable Future. ed. / Maria Chiara Leva; Edoardo Patelli; Luca Podofillini; Simon Wilson. 2022. p. 1852-1858.
Research output: Chapter in book/report/conference proceeding › Conference contribution › Research › peer review
}
TY - GEN
T1 - Uncertainty quantification over spectral density estimation for strong motion process with missing data
AU - Chen, Yu
AU - Patelli, Edoardo
AU - Edwards, Ben
AU - Beer, Michael
AU - Sunny, Jaleena
N1 - Conference code: 32
PY - 2022/8/28
Y1 - 2022/8/28
N2 - In this paper, the challenge of quantifying the uncertainty in the estimation of power spectral density (stationary and nonstationary) of ground motion processes subject to missing data is addressed. Specifically, to exploit additional information besides the incomplete recording, simulated ground motions are generated by a stochastic finitefault model, with its region-specific parameters (source, attenuation, and site parameters) modeled as probability distributions. Then a Bayesian neural network is constructed to probabilistically learn the temporal patterns from such uncertain time-series data. Epistemic uncertainties on the model parameters of the Bayesian neural network model are learnt via variational inference. Thanks to the probabilistic merit of the Bayesian neural network, an ensemble of reconstructed realizations can be obtained, which leads to a probabilistic power spectrum, with each frequency component represented by a probability distribution. This framework is of great importance to researches such as stochastic structural dynamics, where accurate stochastic representations are needed for characterizing engineering excitation processes but faced with incomplete ground motion recordings.
AB - In this paper, the challenge of quantifying the uncertainty in the estimation of power spectral density (stationary and nonstationary) of ground motion processes subject to missing data is addressed. Specifically, to exploit additional information besides the incomplete recording, simulated ground motions are generated by a stochastic finitefault model, with its region-specific parameters (source, attenuation, and site parameters) modeled as probability distributions. Then a Bayesian neural network is constructed to probabilistically learn the temporal patterns from such uncertain time-series data. Epistemic uncertainties on the model parameters of the Bayesian neural network model are learnt via variational inference. Thanks to the probabilistic merit of the Bayesian neural network, an ensemble of reconstructed realizations can be obtained, which leads to a probabilistic power spectrum, with each frequency component represented by a probability distribution. This framework is of great importance to researches such as stochastic structural dynamics, where accurate stochastic representations are needed for characterizing engineering excitation processes but faced with incomplete ground motion recordings.
KW - evolutionary spectral density
KW - ground motion
KW - stochastic processes
KW - uncertainty quantification
KW - variational inference
UR - http://www.scopus.com/inward/record.url?scp=85208234726&partnerID=8YFLogxK
U2 - 10.3850/978-981-18-5183-4_S02-04-389-cd
DO - 10.3850/978-981-18-5183-4_S02-04-389-cd
M3 - Conference contribution
AN - SCOPUS:85208234726
SN - 9789811851834
SP - 1852
EP - 1858
BT - Proceedings of the 32nd European Safety and Reliability Conference, ESREL 2022 - Understanding and Managing Risk and Reliability for a Sustainable Future
A2 - Leva, Maria Chiara
A2 - Patelli, Edoardo
A2 - Podofillini, Luca
A2 - Wilson, Simon
T2 - 32nd European Safety and Reliability Conference (ESREL 2022)
Y2 - 28 August 2022 through 1 September 2022
ER -