Summary
A framework for the generation of bridge‐specific fragility curves utilizing the capabilities of machine learning and stripe‐based approach is presented in this paper. The proposed ...methodology using random forests helps to generate or update fragility curves for a new set of input parameters with less computational effort and expensive resimulation. The methodology does not place any assumptions on the demand model of various components and helps to identify the relative importance of each uncertain variable in their seismic demand model. The methodology is demonstrated through the case study of a multispan concrete bridge class in California. Geometric, material, and structural uncertainties are accounted for in the generation of bridge numerical models and their fragility curves. It is also noted that the traditional lognormality assumption on the demand model leads to unrealistic fragility estimates. Fragility results obtained by the proposed methodology can be deployed in a risk assessment platform such as HAZUS for regional loss estimation.
•Identification of mode of failure of beam-column joints through machine learning techniques.•Probabilistic models to capture the type of failure and shear strength of beam-column joints.•Sensitivity ...of input variables to joint shear strength.•Comparison of various machine learning techniques to estimate the shear strength of beam-column joints.
Beam-column joints are one of critical components that control the oveerall performance of reinforced concrete building frames under seismic loadings. To identify the response mechanism, including the classification of failure mode and the prediction of associated shear strength, of beam-column joints, this paper introduces the application of machine learning techniques. The efficiency of various machine learning techniques is evaluated using extensive experimental data from 536 experimental tests, all of which exhibited either non-ductile joint shear failure prior to beam yielding or ductile joint shear failure after beam yielding. It has been seen from the comparison that lasso regression has a better efficiency and reasonable accuracy in the classification and prediction. The suggested formulations as a function of influential input variables can be easily used by structural engineers to provide an optimal rehabilitation strategy for existing buildings and to design new structures.
Summary
Recent efforts of regional risk assessment of structures often pose a challenge in dealing with the potentially variable uncertain input parameters. The source of uncertainties can be either ...epistemic or aleatoric. This article identifies uncertain variables exhibiting strongest influences on the seismic demand of bridge components through various regression techniques such as linear, stepwise, Ridge, Lasso, and elastic net regressions. The statistical results indicate that Lasso regression is the most effective one in predicting the demand model as it has the lowest mean square error and absolute error. As the sensitivity study identifies more than 1 significant variable, a multiparameter fragility model using Lasso regression is suggested in this paper. The proposed fragility methodology is able to identify the relative impact of each uncertain input variable and level of treatment needed for these variables in the estimation of seismic demand models and fragility curves. Thus, the proposed approach helps bridge owners to spend their resources judiciously (e.g., data collection, field investigations, and censoring) in the generation of a more reliable database for regional risk assessment. This proposed approach can be applicable to other structures.
•Introduce artificial neural network for regional seismic risk assessment of skewed bridges.•Develop multi-dimensional fragilities for California bridges via artificial neural network.•Reduce ...computational efforts for developing bridge-class fragility curves.•Estimate the seismic vulnerability of skewed bridges.
Recent researches are directed towards the regional seismic risk assessment of structures based on a bridge inventory analysis. The framework for traditional regional risk assessments consists of grouping the bridge classes and generating fragility relationships for each bridge class. However, identifying the bridge attributes that dictate the statistically different performances of bridges is often challenging. These attributes also vary depending on the demand parameter under consideration. This paper suggests a multi-parameter fragility methodology using artificial neural network to generate bridge-specific fragility curves without grouping the bridge classes. The proposed methodology helps identify the relative importance of each uncertain parameter on the fragility curves. Results from the case study of skewed box-girder bridges reveal that the ground motion intensity measure, span length, and column longitudinal reinforcement ratio have a significant influence on the seismic fragility of this bridge class.
•Use of Shapely additive explanations for failure modes of RC columns and shear walls.•Importance factor for failure modes of RC columns and shear walls.•Identification of attribute contributions for ...failure mode predictions.•Explanation of the complex machine learning models.•Machine learning-based failure mode prediction models for RC columns and shear walls.
Machine learning approaches can establish the complex and non-linear relationship among input and response variables for the seismic damage assessment of structures. However, lack of explainability of complex machine learning models prevents their use in such assessment. This paper uses extensive experimental databases to suggest random forest machine learning models for failure mode predictions of reinforced concrete columns and shear walls, employs the recently developed SHapley Additive exPlanations approach to rank input variables for identification of failure modes, and explains why the machine learning model predicts a specific failure mode for a given sample or experiment. A random forest model established provides an accuracy of 84% and 86% for unknown data of columns and shear walls, respectively. The geometric variables and reinforcement indices are critical parameters that influence failure modes. The study also reveals that existing strategies of failure mode identification based solely on geometric features are not enough to properly identify failure modes.
To organize accurate and effective emergency responses after an earthquake, it is vital to conduct an early and precise assessment of damage to structures. The use of fragility/vulnerability curves ...is an advanced evaluation approach for structural damage assessments. However, the analysis based on fragility curves significantly varies depending on soil conditions, ground motion, and structural characteristics. To overcome this issue, a stacked long short‐term memory network was proposed in this research. Unlike previous studies, two input features (acceleration time history in the form of vector and the number of stories in the scalar) are utilized to generalize the results for the same plan building frames with different stories. Three different approaches are presented in this work to link the ground motion time history with the number of stories (2, 4, 8, 12, and 20 stories) in the reinforced concrete building frame, and the networks were tested for unknown ground motions. Of the three approaches, those providing good results were selected for further analysis. For the approaches chosen, the network architectures were changed to a diamond shape and an autoencoder‐like shape with more hidden units (to obtain higher accuracy), which were tested for unknown same plan layout frames. The accuracy obtained using these approaches was significantly high (80%–90%) with a low training time. The proposed model is compared with other techniques and shows significant accuracy. The suggested networks exhibited a number of scenarios for estimating the damage state for unknown ground motions, as well as for unknown frames with various stories. Moreover, the capability of the networks to handle more scalar input features is examined by adding them probabilistically; with additional input variables, the networks predicted the damage state with higher accuracy.
•Suggest a data-driven approach for the failure mode prediction of RC shear walls.•Construct an experimental database for RC shear walls.•Compare the performance of prediction models using various ...machine learning techniques.•Identify the critical input parameters affecting the failure mode of shear walls.•Propose the open-source data-driven classification model.
A reinforced concrete shear wall is one of the most critical structural members in buildings, in terms of carrying lateral loads. Despite its importance, post-earthquake reconnaissance and recent experimental studies have highlighted the insufficient safety margins of shear walls. The lack of empirical and mechanics-based models prevents rapid failure mode identification of existing shear walls. This study builds on recent advances in the area of machine learning to determine the failure mode of shear walls as a function of geometric configurations, material properties, and reinforcement details. This study assembles a comprehensive database consisting of 393 experimental results for shear walls with various geometric configurations. Eight machine learning models, including Naïve Bayes, K-Nearest Neighbors, Decision Tree, Random Forest, AdaBoost, XGBoost, LightGBM, and CatBoost were evaluated in this study, in order to establish the best prediction model. As a result of detailed evaluation, a machine learning model based on the Random Forest method is proposed in this paper. The proposed method has 86% accuracy in identifying the failure mode of shear walls. This study also demonstrates that aspect ratio, boundary element reinforcement indices, and wall length-to-wall thickness ratio are the critical parameters influencing the failure mode of shear walls. Finally, an open-source data-driven classification model that can be used in design offices across the world is provided in this paper. The proposed model has the flexibility to account for additional experimental results yielding new insights.
Predicting peak time‐domain ground‐motion parameters, such as peak ground acceleration (PGA), peak ground velocity, and peak ground displacement at a specific location, is challenging because of the ...limited number of recorded ground motions and the complexity of ground‐motion prediction equations. This study presents a novel approach that integrates a geographic information system with a spatial data analysis‐based machine learning PGA prediction model to overcome these challenges and predict PGA classes as a function of the PGA of the respective seismic stations, interstation distance of the seismic stations, and time‐average shear‐wave velocity in the upper 30 m of the target station. The proposed spatial data analysis‐based machine learning approach demonstrated the ability to generate satisfactory results in a short period. To account for the spatial dependencies of the variables, a feature selection method for spatial data using mutual information‐based feature selection was proposed, which provides a well‐prepared spatial matrix for machine learning algorithms. This study evaluates the performance of the model using various machine learning algorithms, including Random Forest, Naïve Bayes, K‐Nearest Neighbors, Decision Tree, AdaptiveBoost, Random Undersampling Boost, Extreme Gradient Boost (XGBoost), and Categorical Boost (CatBoost). Among these, XGBoost and CatBoost performed better than the other methods and yielded fairly accurate results. The models were validated using K‐Fold cross‐validation, and the Wilcoxon signed‐rank test was used for comparison. The spatial data analysis‐based machine learning models, particularly XGBoost and CatBoost, achieved high‐accuracy rates in classifying the PGA levels of 99.1% and 98.9%, respectively. Hyperparameters for the XGBoost model were tuned through GridSearchCV. Tree‐based models outperformed parametric models, indicating complex non‐linear spatial relationships, and by combining spatial feature selection with machine learning models demonstrated improved performance. Additionally, real‐time applications of spatial data analysis‐based machine learning PGA prediction models were used to estimate the seismic vulnerability of postulated concrete box‐girder bridges in San Fernando, thus providing insights into damage probabilities based on predicted PGA values.
This study presents a simple hysteretic model to reproduce the stress–strain relationship of superelastic NiTi shape memory alloys (SMAs). The proposed model explicitly includes the functional ...degradation of SMAs, which has been ignored in earthquake engineering applications. This effect causes a reduction in the transformation stress and accumulation of residual strain. Because SMA devices are mainly used for seismic retrofit and account for a small portion of the structural system, their numerical model should not increase the computational time needed to perform nonlinear dynamic analyses. Computational efficiency can be achieved by representing their stress–strain response in a phenomenological way. Additionally, practitioners who may not have a professional background in materials science can easily manipulate the proposed model for the appropriate reproduction of model parameters such as transformation stress and residual strain. The ability to properly reproduce the experimental stress–strain response is validated for the test results of 65 NiTi SMA specimens. The amount of forward and reverse transformation stress degradation and the amount of residual strain accumulation per cycle, which are observed in the experimental results, are captured with reasonable accuracy in the proposed model. Additionally, the response of SMA braces in a four‐story steel moment frame is modeled using the proposed model to examine the residual story drift of the SMA braced frame under a set of ground motions. At higher intensity levels, the functional degradation of SMA braces increased the residual story drift up to 60% in comparison to the SMA‐braced model without functional degradation.
•Rapid damage assessment of bridges in the transportation networks.•An easy-to-implement machine learning based tagging procedure.•Comparison of various machine learning approaches for damage ...assessment.•Identification of the optimal machine learning procedure for tagging the bridge systems.
The damage state of a bridge has significant implications on the post-earthquake emergency traffic and recovery operations and is critical to identify the post-earthquake damage states without much delay. Currently, the damage states are identified either based on visual inspection or pre-determined fragility curves. Although these methodologies can provide useful information, the timely application of these methodologies for large scale regional damage assessments is often limited due to the manual or computational efforts. This paper proposes a methodology for the rapid damage state assessment (green, yellow, or red) of bridges utilizing the capabilities of machine learning techniques. Contrary to the existing methods, the proposed methodology accounts for bridge-specific attributes in the damage state assessment. The proposed methodology is demonstrated using two-span box-girder bridges in California. The prediction model is established using the training set, and the performance of the model is evaluated using the test set. It is noted that the machine learning algorithm called Random Forest provides better performance for the selected bridges, and its tagging accuracy ranges from 73% to 82% depending on the bridge configuration under consideration. The proposed methodology revealed that input parameters such as span length and reinforcement ratio in addition to the ground motion intensity parameter have a significant influence on the expected damage state.
