Asian Journal of Civil Engineering最新文献

Soil-structure interaction (SSI) under seismic loading is a rather complex phenomenon that has immense effects on the seismic performance of structures. Traditional approaches are the finite element method (FEM) and the boundary element method (BEM), which have been used rather widely in analyzing SSI. Both methods usually fail to capture the complex dynamics of the underlying process. Recent advances in machine learning offer promising alternatives for predictive modeling and analysis of SSI. This paper deals with the applicability of the XGBoost machine learning model, optimized with particle swarm optimization (PSO) in predicting Soil-Structure Interaction under Seismic Loading. The presented model shows accuracy with mean squared error (MSE): 0.04, Root Mean Squared Error (RMSE): 0.2, R-squared (R²): 0.95, and mean absolute error (MAE): 0.1. The results show the better performance of the model over traditional methods like the finite element method (FEM) and the boundary element method (BEM). Comparisons through visualization show that there were close agreements in the displacements predicted and real displacements. Stress distributions and stress–strain curves, predicted from the analysis, validate the model's accuracy. The important outcomes are that the model can deliver more accurate and reliable predictions, enhancing seismic design, and safety to a great extent. It contributes to the literature by being the first application of machine learning combined with an optimization technique; it provides a full comparison to traditional methods for the community and shows future research opportunities, for example, including real-time seismic data or exploring model transferability.

Graphical Abstract

Improving ventilation systems is essential for better indoor air quality, energy efficiency, and overall building performance. This study introduces a new optimization model to tackle the trade-offs between time, cost, and indoor air quality (IAQ) in ventilation system retrofitting projects. Using the Non-dominated Sorting Genetic Algorithm III (NSGA-III), the model evaluates various retrofitting options, including upgrades for ventilation capacity, energy efficiency, air quality, noise reduction, and aesthetic improvements. Each option is assessed for its impact on project duration, cost, and indoor air quality. The goal is to find the best combinations of these options that minimize both project time and cost while improving indoor air quality and meeting resource constraints. The NSGA-III algorithm generates a set of optimal solutions, providing a range of choices for balancing these factors. A comparison with existing methods shows that this new approach offers better solutions for managing these trade-offs. By selecting the most effective solution from these options using a weighted sum method, the study demonstrates NSGA-III’s power in handling complex optimization problems. This model supports better decision-making in retrofitting projects, advancing both sustainability and indoor environment quality.

The construction industry plays a pivotal role in global carbon emissions, prompting a critical need for sustainable infrastructure-development practices. Retaining walls, which are essential for stabilising earth and water pressure in civil engineering projects, represent a significant opportunity to mitigate environmental impacts through material optimisation. This study investigated the design efficiency and embodied carbon and cost implications of cantilever retaining walls constructed with concrete and steel sheet piles. This study employs a thorough methodology that incorporates quantitative studies of the cost and embodied carbon at varying retaining wall heights. The environmental effects and financial viability of the concrete and steel sheet piles were assessed using standardised procedures and local market data. The results indicate that in every height category, concrete sheet piles show consistently reduced total costs and embodied carbon when compared to their steel equivalents. Superior environmental sustainability is demonstrated by concrete, where the embodied carbon levels gradually increase as the wall height increases. On the other hand, steel provides better load-bearing capability, but at a higher cost to the environment and economy, which is especially noticeable in taller structures. This study offers significant perspectives for engineers and other relevant parties to enhance design results that harmonise ecological responsibility with cost-effectiveness in building methods.

Such is the need for this work, since accurate concrete strength prediction at different curing conditions is critical to having structures that are both strong and long-lasting. Traditional methods for the prediction of concrete strength often lack the complexity that occurs in the interaction of the different environmental factors involved, hence leading to suboptimal practices in curing with potential structural weaknesses. Current research into this area has typically focused on disparate data sources and rather naïve modeling methods, further limiting predictive accuracy and creating a general lack of comprehensive knowledge of curing dynamics. Such limitations bring out the need for a more integrated, sophisticated predictive modeling approach to explain variability in concrete strength levels. This paper proposes a novel predictive modeling framework that will be powered by advanced machine learning techniques to take up these challenges. It will adopt a multimodal data integration approach driven by a combination of sensor data related to temperature, humidity, and strain gauges; environmental data related to weather conditions and atmospheric pressure; and historical records, such as mix design and curing duration, further leveraging techniques from data fusion, including the Kalman filter and Bayesian networks. This will be further integrated into a unified, enriched dataset, encapsulating the complex interaction of factors influencing concrete strength. In the present work, this is a chosen approach: hybrid modeling with ensemble learning using XGBoost for the prediction of static features, and Long Short-Term Memory (LSTM) networks for capturing temporal dependencies. In this case, a combination of these models via weighted averaging or stacking improves the accuracy of the predictions to a very great extent: the R² increased from 0.85 to 0.92, and MAE levels by 10–15%. In addition to that, AutoML with Feature Tools implements advanced feature engineering through the generation and selection of optimal features on transformation and aggregation primitives, further refining model performance and interpretability. This process at times reduces the Root Mean Squared Error levels by 5–10%. Finally, Bayesian Neural Networks together with Sobol sensitivity analysis can be used for handling uncertainty and uncovering key factors. BNN provides probabilistic predictions, therefore 95% confidence intervals, while Sobol analysis identifies those critical features that contribute more to variability and allows an in-depth understanding of the role each factor has in driving concrete strength. Indeed, the framework propounded in this work has made great strides in predictive abilities concerning concrete strength variability and will permit more efficient curing practices, thereby making construction outcomes more secure and long-lasting.

This study proposes the hybrid machine learning model combining random forest and Taguchi optimization (RF-TGC) to predict the shear strength of recycled reinforced concrete beams (RARC). For this objective, a total of 128 experimental results of shear strength of RARC beams from published papers were used to develop the proposed RF-TGC model. The performance of the hybrid RF-TGC model was compared with the pure RF model, the k-nearest neighbour (k-NN) model, and the multiple linear regression (MLR) model based on the four indicators of the error metric: mean absolute error (MAE), mean squared error (MSE), root mean squared error (RMSE), and coefficient of determination (R²). As a result, the hybrid RF-TGC model showed the best accuracy in predicting the shear strength of the RARC beam compared to the pure RF, k-NN and MLR models with an R² value of over 0.9 in training and a value of 0.89 in testing. In addition, the sensitivity analyses of the input parameters for the shear strength of the RARC beam were also investigated. It was found that the percentage of transverse steels is the most important parameter for predicting the shear strength of RARC beams. Finally, a free web application was developed to quickly predict the shear strength of the RARC beam in practical implementation.

The paper is a landmark in earthquake and structural engineering, with modern machine-learning techniques applied to introduce innovative investigations into forecasting seismic behavior for vertically uneven structures using sophisticated machine-learning methodologies. The research constructs a very accurate model for making predictions using the XGBoost algorithm with the Owl Search algorithm (OSA) for hyperparameter tuning, which explicitly considers complex behavior in the structures under seismic stresses. The variety within the dataset is broad and covers all kinds of irregularities in the structures, such as stiffness and mass irregularities; thus, it has been used to accurately represent the complex characteristics of actual buildings. The results indicate a strong dependence of base shear capacity and seismic performance on the irregularity of stiffness and mass. The test accuracy of the optimized XGBoost model was 98.8%. The result was better than that of conventional models, thus proving the effectiveness of integrating the Owl Search Algorithm in further fine-tuning the parameters. These results give new variables as insight into affecting earthquake resilience and represent practical applications that enhance building design and retrofitting processes. This is further underlined by the proposal of future research directions that would extend the model’s applicability to other structural anomalies and include additional machine-learning methodologies. Through AI-driven approaches, this study captured complicated structural dynamics with the utmost precision, thus opening new insights that could be brought into practice to improve building design and retrofitting strategies in a way that would diminish the impact of seismic events.

Graphical abstract

Vibration challenges in lightweight pedestrian structures, such as footbridges, have been extensively studied, particularly following the notable lateral vibrations observed during the opening of the London Millennium Bridge on June 10, 2000. This incident underscores the critical need for a deeper understanding of the dynamic behavior of pedestrian bridges subjected to human-induced loads. This study focuses on the dynamic responses of pedestrian steel truss bridges under various loading conditions, including walking, jogging, and crowd-induced vibrations. Finite element analysis is used to identify critical parameters such as the fundamental frequency, acceleration, and damping and evaluate these parameters against the comfort criteria specified in BS EN 1991-2: 2003. Initial findings revealed that acceleration values exceeded the acceptable limits, prompting structural modifications to enhance mass, stiffness, and damping properties. Additionally, incorporating tuned mass dampers as a mitigation strategy demonstrated significant efficacy, achieving up to a 90% reduction in deck acceleration. The results provide valuable insights into optimising pedestrian bridge designs to improve both structural performance and user comfort, contributing to safer and more resilient infrastructures.

In this study, Artificial Neural Networks (ANN) and Response Surface Methodology (RSM) are used to develop models to predict compressive strength, thermal conductivity, porosity and water absorption of eco-friendly fired clay bricks containing different amounts of water, percentages of spent bleaching earth (SBE) and, firing temperatures. Water content was varied between 5 and 8 wt.%, SBE was varied in the range of 0 to 50 wt.% and, firing temperature ranges from 800 to 950 °C. The fired bricks properties were strongly influenced by the SBE content and fairing temperature as confirmed by the SEM images. The percentages of water strongly influenced the compressive strength but had less influence on the porosity and water absorption and no influence on the thermal conductivity. The statistical values for both RSM and ANN models: (coefficient of determination (R²), adjusted coefficient of determination (R² _adj), mean square error (MSE), root mean square error (RMSE) and relative percent deviation (RDP)), were used to compare the two models. The results reveal high correlation coefficients, adjusted coefficients and low root mean square errors. The models were found robust and accurate in their predictions. Based on these results, the RSM and ANN models can be applied as an effective tool to predict compressive strength, thermal conductivity, porosity, and water absorption of fired bricks. Nevertheless, the artificial neural network model showed better accuracy.

Graphical Abstract

This study explores the application of electrical resistivity as a non-destructive method for evaluating concrete properties in reinforced structures. It investigates correlations between surface electrical resistivity (ρ) and fundamental mechanical strengths—compressive (fc), splitting tensile (ft), and flexural (fz) across three concrete grades (M20, M30, M40). Using an Adaptive Neuro-Fuzzy Inference System (ANFIS) in MATLAB, experimental data are analysed to minimize root mean square error (RMSE). The study develops regression models incorporating nonlinear and interaction terms to predict compressive, flexural, and tensile strengths, achieving high coefficients of determination (R² values of 0.94, 0.98, and 0.98 respectively). Validation against experimental data confirms model accuracy, with errors consistently below 10%. This innovative application of ANFIS and electrical resistivity not only enhances the prediction of concrete strengths but also establishes electrical resistivity as a promising tool for non-destructive assessment, crucial for ensuring the structural integrity of concrete infrastructure.