-
Bayesian Structural Model Updating with Multimodal Variational Autoencoder
Authors:
Tatsuya Itoi,
Kazuho Amishiki,
Sangwon Lee,
Taro Yaoyama
Abstract:
A novel framework for Bayesian structural model updating is presented in this study. The proposed method utilizes the surrogate unimodal encoders of a multimodal variational autoencoder (VAE). The method facilitates an approximation of the likelihood when dealing with a small number of observations. It is particularly suitable for high-dimensional correlated simultaneous observations applicable to…
▽ More
A novel framework for Bayesian structural model updating is presented in this study. The proposed method utilizes the surrogate unimodal encoders of a multimodal variational autoencoder (VAE). The method facilitates an approximation of the likelihood when dealing with a small number of observations. It is particularly suitable for high-dimensional correlated simultaneous observations applicable to various dynamic analysis models. The proposed approach was benchmarked using a numerical model of a single-story frame building with acceleration and dynamic strain measurements. Additionally, an example involving a Bayesian update of nonlinear model parameters for a three-degree-of-freedom lumped mass model demonstrates computational efficiency when compared to using the original VAE, while maintaining adequate accuracy for practical applications.
△ Less
Submitted 20 June, 2024; v1 submitted 7 June, 2024;
originally announced June 2024.
-
Site-Specific Ground Motion Generative Model for Crustal Earthquakes in Japan Based on Generative Adversarial Networks
Authors:
Yuma Matsumoto,
Taro Yaoyama,
Sangwon Lee,
Takenori Hida,
Tatsuya Itoi
Abstract:
We develop a site-specific ground-motion model (GMM) for crustal earthquakes in Japan that can directly model the probability distribution of ground motion acceleration time histories based on generative adversarial networks (GANs). The proposed model can generate ground motions conditioned on moment magnitude, rupture distance, and detailed site conditions defined by the average shear-wave veloci…
▽ More
We develop a site-specific ground-motion model (GMM) for crustal earthquakes in Japan that can directly model the probability distribution of ground motion acceleration time histories based on generative adversarial networks (GANs). The proposed model can generate ground motions conditioned on moment magnitude, rupture distance, and detailed site conditions defined by the average shear-wave velocity in the top 5 m, 10 m, and 20 m ($V_{\mathrm{S}5}$, $V_{\mathrm{S}10}$, $V_{\mathrm{S}20}$) and the depth to shear-wave velocities of 1.0 km/s and 1.4 km/s ($Z_{1.0}$, $Z_{1.4}$). We construct the neural networks based on styleGAN2 and introduce a novel neural network architecture to generate ground motions considering the effect of source, path, and such detailed site conditions. 5% damped spectral acceleration of ground motions generated by the proposed GMM is consistent with empirical GMMs in terms of magnitude and distance scaling. The proposed GMM can also generate ground motions accounting for the shear-wave velocity profiles of surface soil with different magnitudes and distances, and represent characteristic that are not explained solely by $V_{\mathrm{S}30}$.
△ Less
Submitted 24 April, 2024;
originally announced April 2024.
-
Latent Space-based Likelihood Estimation Using Single Observation for Bayesian Updating of Nonlinear Hysteretic Model
Authors:
Sangwon Lee,
Taro Yaoyama,
Yuma Matsumoto,
Takenori Hida,
Tatsuya Itoi
Abstract:
This study presents a novel approach to quantifying uncertainties in Bayesian model updating, which is effective in sparse or single observations. Conventional uncertainty quantification metrics such as the Euclidean and Bhattacharyya distance-based metrics are potential in scenarios with ample observations. However, their validation is limited in situations with insufficient data, particularly fo…
▽ More
This study presents a novel approach to quantifying uncertainties in Bayesian model updating, which is effective in sparse or single observations. Conventional uncertainty quantification metrics such as the Euclidean and Bhattacharyya distance-based metrics are potential in scenarios with ample observations. However, their validation is limited in situations with insufficient data, particularly for nonlinear responses like post-yield behavior. Our method addresses this challenge by using the latent space of a Variational Auto-encoder (VAE), a generative model that enables nonparametric likelihood evaluation. This approach is valuable in updating model parameters based on nonlinear seismic responses of structure, wherein data scarcity is a common challenge. Our numerical experiments confirm the ability of the proposed method to accurately update parameters and quantify uncertainties using limited observations. Additionally, these numerical experiments reveal a tendency for increased information about nonlinear behavior to result in decreased uncertainty in terms of estimations. This study provides a robust tool for quantifying uncertainty in scenarios characterized by considerable uncertainty, thereby expanding the applicability of Bayesian updating methods in data-constrained environments.
△ Less
Submitted 4 April, 2024;
originally announced April 2024.
-
Stress-Resultant-Based Approach to Mass-Assumption-Free Bayesian Model Updating of Frame Structures
Authors:
Taro Yaoyama,
Tatsuya Itoi,
Jun Iyama
Abstract:
Bayesian model updating facilitates the calibration of analytical models based on observations and the quantification of uncertainties in model parameters such as stiffness and mass. This process significantly enhances damage assessment and response predictions in existing civil structures. Predominantly, current methods employ modal properties identified from acceleration measurements to evaluate…
▽ More
Bayesian model updating facilitates the calibration of analytical models based on observations and the quantification of uncertainties in model parameters such as stiffness and mass. This process significantly enhances damage assessment and response predictions in existing civil structures. Predominantly, current methods employ modal properties identified from acceleration measurements to evaluate the likelihood of the model parameters. This modal analysis-based likelihood generally involves a prior assumption regarding the mass parameters. In civil structures, accurately determining mass parameters proves challenging owing to the time-varying nature of imposed loads. The resulting inaccuracy potentially introduces biases while estimating the stiffness parameters, which affects the assessment of structural response and associated damage. Addressing this issue, the present study introduces a stress-resultant-based approach for Bayesian model updating independent of mass assumptions. This approach utilizes system identification on strain and acceleration measurements to establish the relationship between nodal displacements and elemental stress resultants. Employing static analysis to depict this relationship aids in assessing the likelihood of stiffness parameters. Integrating this static-analysis-based likelihood with a modal-analysis-based likelihood facilitates the simultaneous estimation of mass and stiffness parameters. The proposed approach was validated using numerical examples on a planar frame and experimental studies on a full-scale moment-resisting steel frame structure.
△ Less
Submitted 31 January, 2024;
originally announced January 2024.
-
Probabilistic Model Updating of Steel Frame Structures Using Strain and Acceleration Measurements: A Multitask Learning Framework
Authors:
Taro Yaoyama,
Tatsuya Itoi,
Jun Iyama
Abstract:
This paper proposes a multitask learning framework for probabilistic model updating by jointly using strain and acceleration measurements. This framework can enhance the structural damage assessment and response prediction of existing steel frame structures with quantified uncertainty. Multitask learning may be used to address multiple similar inference tasks simultaneously to achieve a more robus…
▽ More
This paper proposes a multitask learning framework for probabilistic model updating by jointly using strain and acceleration measurements. This framework can enhance the structural damage assessment and response prediction of existing steel frame structures with quantified uncertainty. Multitask learning may be used to address multiple similar inference tasks simultaneously to achieve a more robust prediction performance by transferring useful knowledge from one task to another, even in situations of data scarcity. In the proposed model-updating procedure, a spatial frame is decomposed into multiple planar frames that are viewed as multiple tasks and jointly analyzed based on the hierarchical Bayesian model, leading to robust estimation results. The procedure uses a displacement-stress relationship in the modal space because it directly reflects the elemental stiffness and requires no prior knowledge concerning the mass, unlike most existing model-updating techniques. Validation of the proposed framework by using a full-scale vibration test on a one-story, one-bay by one-bay moment resisting steel frame, wherein structural damage to the column bases is simulated by loosening the anchor bolts, is presented. The experimental results suggest that the displacement-stress relationship has sufficient sensitivity toward localized damage, and the Bayesian multitask learning approach may result in the efficient use of measurements such that the uncertainty involved in model parameter estimation is reduced. The proposed framework facilitates more robust and informative model updating.
△ Less
Submitted 31 January, 2024;
originally announced January 2024.
-
Polytypes of long-period stacking structures synchronized with chemical order in a dilute Mg-Zn-Y alloy
Authors:
E. Abe,
A. Ono,
T. Itoi,
M. Yamasaki,
Y. Kawamura
Abstract:
A series of structural polytypes formed in an Mg-1at.%Zn-2at.%Y alloy has been identified, which are reasonably viewed as long-period stacking derivatives of the hcp Mg structure with alternate AB stacking of the close-packed atomic layers. Atomic-resolution Z-contrast imaging clearly revealed that the structures are long-period chemical-ordered as well as stacking-ordered; unique chemical order a…
▽ More
A series of structural polytypes formed in an Mg-1at.%Zn-2at.%Y alloy has been identified, which are reasonably viewed as long-period stacking derivatives of the hcp Mg structure with alternate AB stacking of the close-packed atomic layers. Atomic-resolution Z-contrast imaging clearly revealed that the structures are long-period chemical-ordered as well as stacking-ordered; unique chemical order along the stacking direction occurs as being synchronized with a local faulted stacking of AB'C'A, where B' and C' layers are commonly enriched by Zn/Y atoms.
△ Less
Submitted 26 July, 2011; v1 submitted 24 June, 2011;
originally announced June 2011.
