An official website of the United States government
Lateral stiffness of structural components, such as reinforced concrete (RC) columns, plays an important role in resisting the lateral earthquake loads. The lateral stiffness relates the lateral force to the lateral deformation, having a critical effect on the accuracy of the lateral seismic response predictions. The classical methods (e.g. fiber beam–column model) to estimate the lateral stiffness require calculations from section, element, and structural levels, which is time-consuming. Moreover, the shear deformation and bond-slip effect may also need to be included to more accurately calculate the lateral stiffness, which further increases the modeling difficulties and the computational cost. To reduce the computational time and enhance the accuracy of the predictions, this article proposes a novel data-driven method to predict the laterally seismic response based on the estimated lateral stiffness. The proposed method integrates the machine learning (ML) approach with the hysteretic model, where ML is used to compute the parameters that govern the nonlinear properties of the lateral response of target structural components directly from a training set composed of experimental data (i.e. data-driven procedure) and the hysteretic model is used to directly output the lateral stiffness based on the computed parameters and then to perform the seismic analysis. We apply the proposed method to predict the lateral seismic response of various types of RC columns subjected to cyclic loading and ground motions. We present the detailed model formulation for the application, including the developments of a modified hysteretic model, a hybrid optimization algorithm, and two data-driven seismic response solvers. The results predicted by the proposed method are compared with those obtained by classical methods with the experimental data serving as the ground truth, showing that the proposed method significantly outperforms the classical methods in both generalized prediction capabilities and computational efficiency.
more »
« less
This content will become publicly available on January 1, 2026
A real-time structural seismic response prediction framework based on transfer learning and unsupervised learning
Conventional data-driven methods for predicting the seismic response of structures often require extensive data and computational resources. To address these challenges, a novel deep learning framework that can efficiently and accurately predict the structural seismic responses is proposed. The proposed framework overcomes the limitations of the conventional data-driven methods, by utilizing transfer learning based on the most relevant knowledge determined via the unsupervised learning technique. The framework leverages the seismic information history database to identify the most similar previous earthquake, and subsequently transfers the corresponding knowledge from the Structural Seismic Response network (SSR net) to predict structural responses caused by a new earthquake. This innovative method significantly reduces the need for extensive data collection and provides efficient predictions. Case studies demonstrate the framework’s ability to predict seismic structural responses without extensive training or data collection. The framework can reliably capture the complex nonlinear dynamics of structures under seismic loads and offer significant potential for advancing seismic fragility analyses and reliability assessments. Future research will focus on expanding the framework’s applicability to various structural types and further refining its prediction capabilities.
more »
« less
- Award ID(s):
- 1944301
- PAR ID:
- 10628091
- Publisher / Repository:
- Elsevier
- Date Published:
- Journal Name:
- Engineering Structures
- Volume:
- 323
- Issue:
- PA
- ISSN:
- 0141-0296
- Page Range / eLocation ID:
- 119227
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
null (Ed.)Large-scale panel data is ubiquitous in many modern data science applications. Conventional panel data analysis methods fail to address the new challenges, like individual impacts of covariates, endogeneity, embedded low-dimensional structure, and heavy-tailed errors, arising from the innovation of data collection platforms on which applications operate. In response to these challenges, this paper studies large-scale panel data with an interactive effects model. This model takes into account the individual impacts of covariates on each spatial node and removes the exogenous condition by allowing latent factors to affect both covariates and errors. Besides, we waive the sub-Gaussian assumption and allow the errors to be heavy-tailed. Further, we propose a data-driven procedure to learn a parsimonious yet flexible homogeneity structure embedded in high-dimensional individual impacts of covariates. The homogeneity structure assumes that there exists a partition of regression coeffcients where the coeffcients are the same within each group but different between the groups. The homogeneity structure is flexible as it contains many widely assumed low dimensional structures (sparsity, global impact, etc.) as its special cases. Non-asymptotic properties are established to justify the proposed learning procedure. Extensive numerical experiments demonstrate the advantage of the proposed learning procedure over conventional methods especially when the data are generated from heavy-tailed distributions.more » « less
-
The demand for high-performance computing resources has led to a paradigm shift towards massive parallelism using graphics processing units (GPUs) in many scientific disciplines, including machine learning, robotics, quantum chemistry, molecular dynamics, and computational fluid dynamics. In earthquake engineering, artificial intelligence and data-driven methods have gained increasing attention for leveraging GPU-computing for seismic analysis and evaluation for structures and regions. However, in finite-element analysis (FEA) applications for civil structures, the progress in GPU-accelerated simulations has been slower due to the unique challenges of porting structural dynamic analysis to the GPU, including the reliance on different element formulations, nonlinearities, coupled equations of motion, implicit integration schemes, and direct solvers. This research discusses these challenges and potential solutions to fully accelerate the dynamic analysis of civil structural problems. To demonstrate the feasibility of a fully GPU-accelerated FEA framework, a pilot GPU-based program was built for linear-elastic dynamic analyses. In the proposed implementation, the assembly, solver, and response update tasks of FEA were ported to the GPU, while the central-processing unit (CPU) instructed the GPU on how to perform the corresponding computations and off-loaded the simulated response upon completion of the analysis. Since GPU computing is massively parallel, the GPU platform can operate simultaneously on each node and element in the model at once. As a result, finer mesh discretization in FEA will not significantly increase run time on the GPU for the assembly and response update stages. Work remains to refine the program for nonlinear dynamic analysis.more » « less
-
null (Ed.)Since its commissioning in 2004, the UC San Diego Large High-Performance Outdoor Shake Table (LHPOST) has enabled the seismic testing of large structural, geostructural and soil-foundation-structural systems, with its ability to accurately reproduce far- and near-field ground motions. Thirty-four (34) landmark projects were conducted on the LHPOST as a national shared-use equipment facility part of the National Science Foundation (NSF) Network for Earthquake Engineering Simulation (NEES) and currently Natural Hazards Engineering Research Infrastructure (NHERI) programs, and an ISO/IEC Standard 17025:2005 accredited facility. The tallest structures ever tested on a shake table were conducted on the LHPOST, free from height restrictions. Experiments using the LHPOST generate essential knowledge that has greatly advanced seismic design practice and response predictive capabilities for structural, geostructural, and non-structural systems, leading to improved earthquake safety in the community overall. Indeed, the ability to test full-size structures has made it possible to physically validate the seismic performance of various systems that previously could only be studied at reduced scale or with computer models. However, the LHPOST's limitation of 1-DOF (uni-directional) input motion prevented the investigation of important aspects of the seismic response of 3-D structural systems. The LHPOST was originally conceived as a six degrees-of-freedom (6-DOF) shake table but built as a single degree-of-freedom (1-DOF) system due to budget limitations. The LHPOST is currently being upgraded to 6-DOF capabilities. The 6-DOF upgraded LHPOST (LHPOST6) will create a unique, large-scale, high-performance, experimental research facility that will enable research for the advancement of the science, technology, and practice in earthquake engineering. Testing of infrastructure at large scale under realistic multi-DOF seismic excitation is essential to fully understand the seismic response behavior of civil infrastructure systems. The upgraded 6-DOF capabilities will enable the development, calibration, and validation of predictive high-fidelity mathematical/computational models, and verifying effective methods for earthquake disaster mitigation and prevention. Research conducted using the LHPOST6 will improve design codes and construction standards and develop accurate decision-making tools necessary to build and maintain sustainable and disaster-resilient communities. Moreover, it will support the advancement of new and innovative materials, manufacturing methods, detailing, earthquake protective systems, seismic retrofit methods, and construction methods. This paper will provide a brief overview of the 1-DOF LHPOST and the impact of some past landmark projects. It will also describe the upgrade to 6-DOF and the new seismic research and testing that the LHPOST6 facility will enable.more » « less
-
Model parameter updating can enhance the use of nonlinear structural response simulation to guide decision-making in the post-earthquake environment. Since most structures in high seismic regions are not instrumented with sensors, the response history during ground shaking is usually not available after an earthquake. Nevertheless, technologies such as Light Detection and Ranging (LiDAR) and drone-mounted imaging devices have increased the feasibility of measuring residual deformations after the shaking has subsided. It is within this context that a framework for performing nonlinear structural model parameter updating based only on residual drift measurements is proposed. The considered setting is one where a structure is subjected to a sequence of ground motions (without repairs), whereby after each event, the structural model parameters are updated using a Bayesian formulation and the measured residual drift. The methodology is demonstrated by using experimental data from a reinforced concrete bridge pier subjected to six back-to-back ground motions with significant residual drifts recorded after the third, fourth and fifth records in the sequence. The results showed that the updating procedure is able to incrementally (after each record) improve the accuracy of both the concrete and steel model parameters which also enhanced the estimates of the simulated peak and residual drifts.more » « less
