Prediction Of Seismic Response Research Articles

Multi‐input integrative neural network for soil seismic response modeling at KiK‐net downhole array sites

AbstractPrediction of the soil seismic response is of primary importance for geotechnical earthquake engineering. Conventional physics‐based models such as the finite element method (FEM) often face challenges due to inherent model assumptions and uncertainties of model parameters. Furthermore, these physics‐based models require significant computational resources, particularly when simulating seismic responses across numerous soil sites. In this study, a multi‐ input integrative neural network is developed for predicting soil seismic response based on the recorded data from a large number of downhole array sites. Ground motions, seismic event information, and wave velocity structures of the sites are utilized as input data in the proposed neural network, enabling the model to adapt to various site conditions. Comparative assessments against state‐of‐the‐art FEM models demonstrate that the proposed models exhibit superior prediction performance with increased efficiency. Furthermore, the pre‐training technique, a transfer learning method, is employed to predict the seismic response at new stations. By fine‐tuning the pre‐trained model derived from the extensive dataset with limited recorded data from new stations, high‐precision seismic response predictions can be realized, illustrating the adaptability and efficacy of the proposed approach in data‐scarce conditions.

Seismic response prediction and parameters estimation of the frame structure equipped with the base isolation-fluid inerter system (FS-BIFI) based on the PI-LSTM model

This paper introduces a novel approach, the physics-informed long short-term memory (PI-LSTM) model, to address the forward and inverse problems of the frame structure equipped with the base isolation-fluid inerter system (FS-BIFI). Validation of the PI-LSTM model's effectiveness is demonstrated through a numerical case study of a single-story FS-BIFI. Employing this approach, the PI-LSTM model accurately predicted displacement and acceleration responses of a three-story FS-BIFI. Moreover, it conducts an evaluation of unknown damping-related parameters (β1, β2) and a stiffness-related parameter (μ) of the fluid inerter mounted on the FS-BIFI. The model demonstrates a robust confidence value of 96.68% in predicting response error values within the range of [− 10%, 10%]. Notably, the relative errors in estimating unknown parameters (β1, β2, μ) stand at 12.760%, 8.857%, and 3.750%, respectively. The results show that the PI-LSTM model has great potential for response prediction and parameter inversion of complex structural systems.

High-speed railway seismic response prediction using CNN-LSTM hybrid neural network

High-speed railway seismic response prediction using CNN-LSTM hybrid neural network

Prior knowledge‐infused neural network for efficient performance assessment of structures through few‐shot incremental learning

AbstractStructural seismic safety assessment is a critical task in maintaining the resilience of existing civil and infrastructures. This task commonly requires accurate predictions of structural responses under stochastic intensive ground accelerations via time‐costly numerical simulations. While numerous studies have attempted to use machine learning (ML) techniques as surrogate models to alleviate this computing burden, a large number of numerical simulations are still required for training ML models. Therefore, this study proposes a prior knowledge‐infused neural network (PKNN) for achieving efficient structural seismic response predictions and seismic safety assessment at a low computation cost. In this approach, first, the prior knowledge inherently within a theoretical dynamic model of a structure is infused into a neural network. Then, by utilizing the few‐shot incremental learning technique, this network would be further fine‐tuned by only a few numerical simulations. The resulting PKNN would be able to accurately predict the seismic response of a structure and facilitate seismic safety assessment. In this study, the PKNN's accuracy and computational efficiency are validated on a typical frame structure. The results revealed that the proposed PKNN could be used for accurately predicting the structural seismic responses and assessing the structural safety under a low computational cost.

Predictions of seismic response for sheet-pile wall in liquefiable deposit within LEAP 2022 using the CycLiq constitutive model

The Liquefaction Experiments and Analysis Projects (LEAP) organized centrifuge tests and corresponding numerical predictions for the seismic response of sheet-pile wall in liquefiable deposits in 2022. Within LEAP 2022, two versions of the CycLiq constitutive model, i.e., the original (CycLiq) and modified (CycLiq-M) versions, were calibrated and used for numerical predictions. CycLiq-M was first calibrated using cyclic direct simple shear (CDSS) element tests on the test material, Ottawa F65 sand, and then used for Type-A predictions of another set of CDSS tests for which the results were priorly unknown, validating the model's performance at the element level. Type-B predictions of centrifuge shaking table model tests on sheet-pile wall in liquefiable deposit was then performed using both versions of the calibrated model, without prior knowledge of test results. The results show that the properly calibrated CycLiq model can make reasonable predictions for soil acceleration, excess pore pressure, settlement, and especially sheet-pile wall displacement. In particular, the CycLiq-M version with enhanced cyclic resistance simulation capability exhibits significant improvement in prediction accuracy. Using numerical simulation, the influence of soil-container friction in centrifuge model tests was assessed.

Physics knowledge-based transfer learning between buildings for seismic response prediction

The recent advance in deep learning has attracted considerable interest for employing the state-of-the-art methods to solve engineering problems. However, the applicability of machine learning based models is hindered by the high cost of big data acquisition and task-specific difficulties. This paper presents a framework of physics knowledge-based transfer learning (Phy-KTL) neural networks that integrates the powerful learning capacity of physics-informed neural networks (PINNs) and the flexible transferability of model-based transfer learning technique to enhance structural seismic response prediction in the context of limited labelled datasets. The leverage of physics knowledge (represented by Runge-Kutta solver) allows the neural networks to better capture the structural nonlinear pattern. The use of model-based transfer learning improves the model generality by transferring features extracted from the source building to target buildings. The effectiveness of Phy-KTL in predicting seismic responses between target buildings is numerically validated as compared with Data-driven neural networks, PINNs, and Data-based transfer learning (Data-KTL). A practical application, which uses Phy-KTL to transfer features extracted from the numerical model to the physical building tested on the shaking table, validates that Phy-KTL is robust and effective to improve seismic response prediction of target buildings with limited labelled data.

Seismic response prediction using a hybrid unsupervised and supervised machine learning in case of 3D RC frame buildings

The seismic vulnerability assessment represents an important step in monitoring the buildings’ capacity and checking their performance during and after earthquake events. The Nonlinear Time History Analysis (NL-THA) is considered the most reliable method that is used to calculate the exact structural behavior of any building. However, this sophisticated method is known for its complexity, the use of Finite Element (FE) software, and computational time consuming, especially in the case of tall buildings. For that reason, An Artificial Neural network (ANN) is used to develop a new model able to predict the essential Engineering Demand Parameters (EDPs), i.e., the Maximum Base Shear (MBS), the Maximum Inter-story Drift (MIDR) and the Maximum Roof Drift Ratio (RDR). Unsupervised algorithms such as the Principal Component Analysis PCA and the Autoencoder are coupled with the ANN to reduce the dimensionality, improve the dataset quality, and reduce the irrelevant features. More than 192,000 buildings are analyzed using the NL-THA and eighty artificial ground motions (GMs) to generate the dataset. The buildings’ characteristics are generated randomly from the selection range. The results showed that the Autoencoder-ANN model represents the highest performance compared to the PCA-ANN and ANN models. The Autoencoder-ANN model could quickly and accurately predict the seismic responses of unseen ground motions using only the building’s characteristics and the GM parameters without using any FE software.

Physics-informed artificial intelligence models for the seismic response prediction of rocking structures

Abstract The seismic response of a wide variety of structures, from small but irreplaceable museum exhibits to large bridge systems, is characterized by rocking. In addition, rocking motion is increasingly being used as a seismic protective strategy to limit the amount of seismic actions (moments) developed at the base of structures. However, rocking is a highly nonlinear phenomenon governed by non-smooth dynamic phases that make its prediction difficult. This study presents an alternative approach to rocking estimation based on a physics-informed convolutional neural network (PICNN). By training a group of PICNNs using limited datasets obtained from numerical simulations and encoding the known physics into the PICNNs, important predictive benefits are obtained relieving difficulties associated with over-fitting and minimizing the requirement for a large training database. Two models are created depending on the validation of the deep PICNN: the first model assumes that state variables including rotations and angular velocities are available, while the second model is useful when only acceleration measurements are known. The analysis is initiated by implementing K-means clustering. This is followed by a detailed statistical assessment and a comparative analysis of the response-histories of a rocking block. It is observed that the deep PICNN is capable of effectively estimating the seismic rocking response history when the rigid block does not overturn.

Adaptive Conformal Prediction Intervals Using Data-Dependent Weights with Application to Seismic Response Prediction

Adaptive Conformal Prediction Intervals Using Data-Dependent Weights with Application to Seismic Response Prediction

Seismic response prediction of a damped structure based on data-driven machine learning methods

Damping technology has been widely used because of its good vibration control effect. However, due to the strong nonlinearity of the added dampers, accurately predicting the seismic response of damped structures remains a challenge. This study investigates the application of interpretable machine learning (ML)-based and deep learning-based approaches to the prediction of the maximum inter-storey displacement of a damped structure. A comprehensive database consisting of 13,855 structural responses to ground motions was collected. Seven traditional interpretable ML algorithms including random forest and extreme gradient boosting (XGBoost), a convolutional neural network based on large receptive field, and seismic wave transformer (SWT) model based on a transformer network were developed. The predictions show that the error of the SWT based on unsupervised feature extraction is reduced by 50.90% compared with that of the optimal XGBoost in ensemble learning. Although the SWT has the highest global accuracy, XGBoost is found to have a smaller error when the structure is in a linear state with peak ground acceleration as partition index, so an aggregation model (AM)-based structural response prediction method was also proposed. The accuracy of the AM improved by 27.95% compared with that of the SWT. In contrast with other ML models, the proposed AM is more advantageous in terms of computational efficiency and accuracy.

A rapid analysis framework for seismic response prediction and running safety assessment of train-bridge coupled systems

With high-speed railway lines increasing gradually, the running safety assessment (RSA) of train-bridge coupled (TBC) systems has become an indispensable part of railway seismic design. However, the modeling of TBC systems is difficult and computationally complex. Therefore, it is difficult to conduct real-time assessment and response analysis rapidly. With this in mind, this paper integrates a surrogate framework based on deep learning methods and can help rapid-diagnose the seismic response and RSA. Meanwhile, real-time information feedback is performed to determine the safe speed of the train. To this end, a convolutional neural network-long short-term memory (CNN-LSTM) model with excellent time-series data processing capability is newly used for the seismic response analysis of the TBC system. Furthermore, response data obtained are processed and input to an interpretable adaptive network-based fuzzy inference system (ANFIS). The ANFIS derives a series of fuzzy rules while conducting the RSA, and these rules can help non-specialists understand the assessment process. Finally, parameter inversion is performed by the ANFIS to determine a safe threshold of train speed. According to the tests, the proposed rapid analysis framework can be applied to seismic response analysis and RSA with high efficiency and accuracy in multiple engineering scenarios. The framework may be helpful for railway design and make it possible to realize real-time data analysis for TBC systems.

Seismic loss assessment of code-compliant moment-resisting RC buildings located on different soil conditions of Mexico City

This research aims to estimate the expected economic losses associated with the repair cost of a set of code-compliant moment-resisting reinforced concrete buildings located on different soil conditions in Mexico City. The loss assessment methodology is based on the second generation of the Performance-Based Earthquake Engineering (PBEE) framework, which quantifies the seismic performance of buildings in terms of metrics such as economic losses, downtime, and casualties, which are more meaningful to owners and stakeholders for the decision-making process. The methodology uses a probabilistic approach that takes into account uncertainties in seismic intensity, structural response, component damage, and consequence prediction. The seismic response of structures was calculated in terms of inter-story drifts and floor accelerations to assess the damage to their structural and non-structural components. In addition, taking into account the seismic hazard of the site, the expected annual losses (EAL) are calculated to provide insight into the seismic performance evaluation of structures with different characteristics in terms of the financial impact in seismic-prone regions like Mexico City.

Rapid peak seismic response prediction of two-story and three-span subway stations using deep learning method

A deep learning-based rapid peak seismic response prediction model for the most common two-story and three-span subway stations is proposed in this study. The established model predicts the peak seismic responses of subway stations with a data-driven fashion and using limited information. The prediction model extracts the features of ground motions using one-dimensional convolutional neural network (1D-CNN) and then integrates the information of subway stations (i.e., the seismic fortification intensity, buried depth, and shear wave velocity) through a fully connected neural network for regression, resulting in peak seismic responses, namely the peak floor acceleration (PFA) and maximum inter-story drift ratio (MIDR). The model is trained using 19,200 samples obtained from the nonlinear time-history analyses (NLTHAs) of the designed 48 typical subway station structures. Furthermore, the external model verification was performed on 960 additional samples. For the predictions of PFA and MIDR, the coefficient of determination (R2) values are 0.967 and 0.986, respectively, and the damage states of subway stations are further evaluated, achieving an accuracy of 95.0%. These indicates that the model has good predictive performance and generalization ability. Moreover, the prediction model demonstrates a significantly higher computational efficiency compared to numerical simulation methods.

Physics-informed neural networks for enhancing structural seismic response prediction with pseudo-labelling

Physics-informed neural networks for enhancing structural seismic response prediction with pseudo-labelling

Physics-informed deep 1D CNN compiled in extended state space fusion for seismic response modeling

Artificial neural networks have been proven promisingly powerful in developing a data-driven surrogate model for rapid seismic response modeling, while very few of them embody a physical mechanism with explicit interpretability. Herewith, this study proposes a novel physics-informed deep 1D convolutional neural network compiled in extended state space fusion (SSM-CNN) for enhanced seismic response modeling. In the SSM-CNN, an innovative parameter-free physics-constrained mechanism is designed and embedded for performance enhancement by construing the differential nexus of state variables derived from the state-space representation of initial structural response. Both the designing philosophy and mechanism translating of the SSM-CNN are elaborated exhaustively. To fully demonstrate the capability and superiority of proposed model compared to normal CNN, two groups of filed records of earthquake events with different sources from PEER and CESMD dataset were allocated for numerical and experimental validations, respectively. The validation results confirmed the effectiveness and superiority of physics-informed SSM-CNN in seismic response prediction. The sensitivity analysis and generalization analysis were also performed on the models for better understanding their interpretability. Involved algorithms and models in present study were integrated and developed into a web program, which affirmed a promising engineering applicability for rapid seismic response prediction.

Seismic response and performance prediction of steel buckling-restrained braced frames using machine-learning methods

Nowadays, Buckling-Restrained Brace Frames (BRBFs) have been used as lateral force-resisting systems for low-, to mid-rise buildings. Residual Interstory Drift (RID) of BRBFs plays a key role in deciding to retrofit buildings after seismic excitation; however, existing formulas have limitations and cannot effectively help civil engineers, e.g., FEMA P-58, which is a conservative estimation method. Therefore, there is a need to provide a comprehensive tool for estimating seismic responses of Interstory Drift (ID) and RID with novel approaches to fulfill the shortcomings of existing formulas. The Machine Learning (ML) method is an interdisciplinary approach that makes it possible to solve these types of engineering problems. Therefore, the current study proposes ML algorithms to provide a prediction model for determining the seismic response, seismic performance curve, and seismic failure probability curve of BRBFs. To train ML-based prediction models, Nonlinear Time-History Analysis (NTHA) and Incremental Dynamic Analysis (IDA) were performed on the 2-, to 12-Story BRBFs subjected to 78 far-field ground motions, and 606944 data points were prepared for different prediction purposes. The results indicate that the considered approaches are justified. For instance, the proposed ML methods have the ability to predict the maximum ID, maximum RID and maximum roof ID with the accuracy of even 98.7%, 95.2%, and 93.8%, respectively, for the 4-Story BRBF. Moreover, a general preliminary estimation tool is introduced to provide predictions based on the input parameters considered in the study.

A Pre-Trained Deep Learning Model for Fast Online Prediction of Structural Seismic Responses

Deep learning techniques have gradually attracted considerable research interest in numerous application scenarios because of their capacity to simplify and accelerate calculations. Several researchers have adopted deep learning models, primarily end-to-end long short-term memory networks, to predict structural seismic responses in a data-driven manner and have achieved remarkable improvements. However, further research is required to reduce the training cost and complexity while maintaining the prediction accuracy of end-to-end models. In this study, a pre-training strategy was developed to meet these requirements. In brief, a generic base model was pre-trained by embedding the knowledge of calculating the response of single-degree-of-freedom systems, which was used as the basis for building tailored top models for the online prediction of seismic responses of different structures. This strategy modularized and simplified deep-learning-based seismic response prediction for training while maintaining the advantages of the end-to-end method. Proposed method was verified by using both simulation and actual datasets. The results demonstrated that a significantly lower training cost and complexity were required to achieve a prediction accuracy similar to that of the classic end-to-end method. Specifically, there was a reduction in FLOPs ranging from 20% to 80%, and a decrease in training time ranging from 44% to 82%. Moreover, the performance of deep learning with limited actual data was discussed.

Structural seismic response prediction based on convolutional neural networks

The seismic response of buildings is crucial for structural performance analysis. For structures with complete design data, the seismic response can be predicted using finite element analysis. However, for structures lacking necessary information, building finite element models and predicting their seismic response can be challenging. Compared to finite element analysis, convolutional neural networks (CNNs) can establish a neural network mapping relationship between the structure and the seismic response to predict the structural response without design data. In this paper, a structural response prediction model based on CNNs is established, aiming to analyze the effect of natural frequency reduction on the structural response after the Tohoku earthquake. The successful prediction of the structural acceleration and displacement response provides a new analytical method for predicting the seismic response of buildings lacking design data.

Regional-scale nonlinear structural seismic response prediction by neural network

Urban seismic damage assessment has recently become an emerging research topic due to the accelerating global urbanization trend, to which the seismic responses of building clusters to various earthquakes are a prerequisite. Traditional methods for this task, including vulnerability analysis and time-consuming time history analysis, may suffer from accuracy or efficiency problems especially for nonlinear response calculation. Machine learning methods allow for rapid and accurate response prediction, but current applications still lack scalability (on the size of structures or earthquakes) and the corresponding real datasets. To tackle this issue, this paper proposes an artificial neural network framework for simultaneously predicting nonlinear seismic responses of all buildings in a cluster subjected to multi-earthquake inputs. Inspired by the advanced collaborative filtering techniques, the framework converts the regional response prediction into a matrix completion problem, thereby aggregating information extracted from historical response records and physical characteristics to improve performance. The framework is used to assess nonlinear responses of a real urban region consisting of 2788 buildings subjected to 3798 measured ground motions. The results clearly demonstrated that the proposed framework achieves orders of magnitude faster than time history analysis and average errors below 3 % on several response metrics, showing high computational efficiency and accuracy.

A mode selection procedure for a seismic response prediction method based on microtremor measurements

A novel mode selection procedure is proposed for a seismic response prediction method based on microtremor measurements. The prediction method utilizes modal properties of limited modes to model the objective building. The modal properties are determined through microtremor measurements, and hence, modeling does not require design documents. The prediction method utilizes the identified mode shapes to approximate the corresponding participation factors. However, the approximation results might significantly vary after incorporating a new mode. Accordingly, if the incorporated mode is highly linearly relevant to the previously used modes, the approximation accuracy sharply decreases. This results in a less accurate prediction of seismic responses. Therefore, this paper proposes a mode selection procedure for the prediction method. The effectiveness of the procedure is validated through a numerical simulation. Further, the simulation results indicate that using all known modes in the prediction method might significantly decrease the prediction accuracy compared with using selected modes.