Machine learning (ML) has evolved rapidly over recent years with the promise to substantially alter and enhance the role of data science in a variety of disciplines. Compared with traditional ...approaches, ML offers advantages to handle complex problems, provide computational efficiency, propagate and treat uncertainties, and facilitate decision making. Also, the maturing of ML has led to significant advances in not only the main-stream artificial intelligence (AI) research but also other science and engineering fields, such as material science, bioengineering, construction management, and transportation engineering. This study conducts a comprehensive review of the progress and challenges of implementing ML in the earthquake engineering domain. A hierarchical attribute matrix is adopted to categorize the existing literature based on four traits identified in the field, such as ML method, topic area, data resource, and scale of analysis. The state-of-the-art review indicates to what extent ML has been applied in four topic areas of earthquake engineering, including seismic hazard analysis, system identification and damage detection, seismic fragility assessment, and structural control for earthquake mitigation. Moreover, research challenges and the associated future research needs are discussed, which include embracing the next generation of data sharing and sensor technologies, implementing more advanced ML techniques, and developing physics-guided ML models.
This study conducts a comprehensive review of the promises and challenges of implementing Bayesian statistics in natural hazards engineering. The reviewed natural hazards include earthquakes, floods, ...extreme wind events, wildfires, and landslides and debris flows. An attributes matrix is developed under each hazard to analyze each study based on its associated scale of analysis, topic area, Bayesian method, and data resource. In particular, the state-of-the-art survey elaborates the level of involvement for three categories of Bayesian methods, such as Bayesian model updating, Bayesian network, and Bayesian neural network, in the topic areas of hazard analysis, risk assessment, and structural health monitoring. In general, the existing research in natural hazards engineering is benefited by leveraging Bayesian statistics to handle uncertainties explicitly and deal with large-scale problems that involve different types of data inputs. However, the substantial computational cost and the determination of prior probability distributions are two major challenges bottlenecking the future development of Bayesian statistics. Compared with machine learning, Bayesian approaches offer more transparent model inference and exhibit different abilities to avoid data over fitting. This reviewed work can serve as a sound reference for interested practitioners and researchers to practice, develop, and promote broader and more in-depth Bayesian advances in solving grand challenges in natural hazards engineering.
AbstractDesign of seismic protective devices for highway bridges is often a highly iterative and tedious process due to the nonlinear behavior of the system, a large range of design parameters, and ...uncertainty of ground motions. This paper develops a design surface of system-level repair cost ratios (i.e., the repair costs normalized by bridge replacement costs) for various base-isolated bridges to facilitate the performance-based design and optimization of seismic protective devices. First, component-level fragility functions are derived for multiple base-isolation design cases as functions of earthquake input intensity. Second, nearly identical fragility functions for different isolation designs are obtained when failure probabilities are conditioned on the median engineering demand parameters (EDPs) instead. Subsequently, system-level repair cost ratio is derived by combining the failure probabilities of bridge components in terms of various EDPs. The function surface of the derived repair cost ratio with respect to EDPs is identical across different isolation designs. Hence it can serve as a performance index to facilitate the design and optimization of seismic protective devices. The proposed framework is demonstrated through a case study on Painter Street Overcrossing, a typical highway bridge in California. It is shown that optimal design parameters can be obtained to significantly reduce the overall repair cost ratio of the bridge with consideration of uncertainties and variabilities of ground motions.
Most regional seismic damage assessment (RSDA) methods are based on the rigid-base assumption to ensure evaluating efficiency, while these practices introduce factual errors due to neglecting the ...soil–structure interaction (SSI). Predicting the influence of the SSI on seismic responses of regionwide structure portfolios remains a challenging undertaking, as it requires developing numerous high-fidelity, integrated models to capture the dynamic interplay and uncertainties in structures, foundations, and supporting soils. This study develops a one-dimensional convolutional neural network (1D-CNN) model to efficiently predict to what degree considering the SSI would change the inter-story drifts and base shear forces of RC frame buildings. An experimentally validated finite element model is developed to simulate the nonlinear seismic behavior of the building-foundation–soil system. Subsequently, a database comprising input data (i.e., structural and soil parameters, ground motions) and output predictors (i.e., changes in story drift and base shear) is constructed by simulating 1380 pairs of fixed-base versus soil-supported structures under earthquake loading. This large-scale dataset is used to train, test, and identify the optimal hyperparameters for the 1D-CNN model to quantify the demand differences in inter-story drifts and base shears due to the SSI. Results indicate the 1D-CNN model has a superior performance, and the absolute prediction errors of the SSI influence coefficients for the maximum base shear and inter-story drift are within 9.3% and 11.7% for 80% of cases in the testing set. The deep learning model can be conveniently applied to enhance the accuracy of the RSDA of RC buildings by updating their seismic responses where no SSI is considered.
•p-y spring approach accurately captures multiple SSI effects for highway bridges.•Statistical regressions are developed to identify critical SSI modeling parameters.•Columns and deck are less ...sensitive to SSI modeling for the bridges studied.•Careful soil evaluation is needed for damage estimate of foundations and abutments.
This study investigates the sensitivity of seismic demands and fragility estimates of a typical highway bridge in California to variation in its soil-structure interaction (SSI) modeling parameters. A rigorous p-y spring based modeling approach is developed and validated for an instrumented highway overcrossing that provides a dependable screening of each modeling parameter. Modifications are made to benchmark the overcrossing against typical bridge designs in California, including the consideration of diaphragm and seat abutments. Plausible variation in SSI modeling parameters is established using 18 random variables that cover different soil zones. Influential SSI parameters are identified for the seismic demands of bridge components through two regression techniques such as stepwise and LASSO regressions. Concurring results from both regressions indicate that bridge demand models tend to be sensitive to the modeling parameters associated with near-ground soils. Furthermore, the relative importance of the uncertainty in SSI modeling parameters is assessed with respect to the fragility estimates in both component and system levels. The study reveals that the bridge performance and fragility curves of bridge columns and decks are dominated by the uncertainty in the ground motion. However, the propagation of the potentially variable SSI parameters plays a significant role in the fragility estimates of bridge foundations and abutment components such as span unseating, bearings and shear keys. The results offer insights to guide future uncertainty treatment in SSI modeling and investment in refined soil parameter estimates through field testing or other measures.
AbstractBase isolators and fluid viscous dampers are viable protective devices that have been commonly considered in the seismic protection of civil engineering structures. However, the optimal ...design of these devices remains a tedious and iterative undertaking due to the uncertainty of ground motions, the nonlinear behavior of the structure, and its change of dynamic characteristics (i.e., effective stiffness and damping ratio) under each new design. The optimal design problem becomes more challenging concerning a multiresponse bridge system where conflicting damage potential is often expected among multiple bridge components (e.g., column, bearing, shear key, deck unseating, foundation). In this respect, this study develops a risk-based optimization strategy that directly links the expected annual repair cost ratio (ARCR) of the bridge to the design parameters of base isolators and fluid dampers. This strategy is achieved by devising a multistep workflow that integrates a seismic hazard model, a design of experiment for bearings and dampers, a logistic regression towards parameterized component-level fragility models, and a bridge system-level seismic loss assessment. The developed ARCR is parameterized as a convex function of the influential parameters of seismic protective devices. As such, optimal bearing and damper designs can be pinpointed by directly visualizing the global minimum of the parameterized ARCR surface. The optimal design is carried out against a typical reinforced concrete highway bridge in California that is installed with the fluid dampers and three types of widely-used isolation bearings—the elastomeric bearing, lead-rubber bearing, and friction pendulum system. It is shown that optimal design parameters can be obtained to significantly reduce the expected ARCR of the bridge, whereas combining optimally designed bearings and dampers can provide the minimum seismic risk.
•Numerical model predicts column behavior under long-duration ground motions.•The Park and Ang damage index is developed to capture the cumulative seismic damage.•The bridge column/system show ...elevated seismic fragility under CSEs at higher damage states.•The ARCR and ART of the bridge class are higher when subjected to CSEs versus CEs and SCEs.
This study conducts seismic risk assessment of highway bridges in western Canada. The performance-based earthquake engineering (PBEE) framework is enhanced to assess the expected annual repair cost ratio (ARCR) and annual restoration time (ART) of a benchmark bridge class under the region’s three types of earthquakes - shallow crustal earthquakes (CEs), deep subcrustal earthquakes (SCEs), and megathrust Cascadia subduction earthquakes (CSEs). First, event-specific seismic hazard models are considered, whereas event-consistent ground motions are selected for non-linear time history analyses. Compared with those from CEs and SCEs, CSE ground motions feature a much longer duration. This long-duration effect is captured by validating the numerical model of the bridge column against (1) a cyclic pushover test under standard versus long-duration loading protocols and (2) a shaking table test excited by six consecutive ground motions. Besides, the Park and Ang damage index is utilized as the column’s engineering demand parameter (EDP) and updated as a demand-capacity ratio model when reaching four different damage states. A comprehensive list of ground motion intensity measures (IMs) is considered where the spectra acceleration at one second, Sa(1.0), is chosen as the most suitable IM based on its performance in proficiency, efficiency, practicality, and EDP-IM correlation across all three earthquake events. Subsequently, component- and system-level fragility models are derived under each earthquake type using the cloud analysis that convolves the seismic demands with capacity models for multiple bridge components. To further quantify and propagate the epistemic uncertainty associated with the development of probabilistic seismic demand models (PSDMs), the bootstrap resampling technique is utilized to generate numerous seismic demand datasets and develop a stochastic set of seismic fragility curves. Finally, the bootstrapped, event-dependent fragility models are combined with the respective hazard models and probabilistic loss functions to assess the expected ARCR and ART for the benchmark bridge class. This study underscores the significantly higher seismic risk of highway bridges when facing CSEs, followed by CEs and SCEs.
This study develops an end‐to‐end deep learning framework to learn and analyze ground motions (GMs) through their latent features, and achieve reliable GM classification, selection, and generation of ...simulated motions. The framework is composed of an analysis workflow that transforms and reconstructs GMs through short‐time Fourier transform (STFT), encodes and decodes their latent features through convolutional variational autoencoder (CVAE), and classifies and generates GMs by grouping and interpolating latent variables. A benchmark study is established to confirm the minor difference between original GMs and the corresponding reconstructed accelerograms. The encoded latent space reveals that certain latent variables are directly linked to the dominant physical features of GMs. Resultantly, clustering latent variables using the k‐means algorithm successfully classifies GMs into different groups that vary in earthquake magnitude, soil type, field distance, and fault mechanism. By linearly interpolating two parent latent variables, simulated GMs are generated with consistent class information and matching response spectra. Furthermore, seismic fragility models are developed for a steel frame building and a concrete bridge using different sets of GMs. Using five classified, top‐ranked motions, regardless of recorded or simulated accelerograms, can achieve reasonable and efficient fragility estimates compared to the case that adopts 230 GMs. The proposed deep learning framework addresses two compelling questions regarding seismic fragility assessment: How many GMs are sufficient and what types of motions should be selected.
Alzheimer's disease (AD) is one of the most common neurodegenerative causes of dementia, the pathology of which is still not much clear. It′s challenging to discover the disease modifying agents for ...the prevention and treatment of AD over the years. Emerging evidence has been accumulated to reveal the crucial role of up‐regulated glutaminyl cyclase (QC) in the initiation of AD. In the current study, the QC inhibitory potency of a library consisting of 1621 FDA‐approved compounds was assessed. A total of 54 hits, 3.33 % of the pool, exhibited QC inhibitory activities. The Ki of the top 5 compounds with the highest QC inhibitory activities were measured. Among these selected hits, compounds affecting neuronal signaling pathways and other mechanisms were recognized. Moreover, several polyphenol derivatives with QC inhibitory activities were also identified. Frameworks and subsets contained in these hits were analyzed. Taken together, our results may contribute to the discovery and development of novel QC inhibitors as potential anti‐AD agents.
Up‐regulated glutaminyl cyclase (QC) has a crucial role in the initiation of Alzheimer's disease (AD).The QC inhibitory potency of a library containing 1621 FDA‐approved compounds was investigated. Molecule scaffolds and motifs contained in these hits (3.33 % of the pool) are suggested to be combined into the design of new fragment like libraries targeting QC. This approach may support the development of novel QC inhibitors as potential anti‐AD agents.