AI in civil engineering最新文献

Recent advancements in automated tunnel defect detection have utilized high-resolution cameras and mobile laser scanners. However, the inability of cameras to accurately capture 3D spatial coordinates complicates tasks such as 3D visualization, while the relatively low resolution of laser scanners makes it difficult to detect small defects such as microcracks. In this paper, a comprehensive inspection method is proposed to address these limitations by integrating multi-defect detection, 3D coordinate acquisition, and visualization. The inspection process involves the capture of both image data and point cloud data of tunnel linings using the newly developed inspection cart (MTI-300). The proposed fusion approach combines image and point cloud data, leveraging the enhanced YOLOv8-seg instance segmentation model for defect identification. The scale-invariant feature transform (SIFT) algorithm is used to match local defect regions in the image data with the corresponding point cloud data, enabling the extraction of 3D coordinates and the integration of defect pixels with the point cloud information. Subsequently, a lightweight 3D reconstruction model is developed to visualize the entire tunnel and its defects using the fused data. The performance of the proposed method is validated and substantiated through a field experiment on Metro Line 8 in Qingdao, China.

Strip foundations, as a widely applied form of shallow foundation, involve foundation displacements and soil deformations under loading, which are critical issues in geotechnical engineering. Traditional limit analysis methods can only provide solutions for ultimate bearing capacity, while numerical methods require remeshing and remodeling for different scenarios. To address these challenges, this study proposes a deep learning approach based on the DeepONet neural operator for rapid and accurate predictions of load–displacement curves and vertical displacement fields of strip foundations under various conditions. A dataset with randomly distributed parameters was generated using finite element method, with the training set employed to train the neural network. Validation on the test set shows that the proposed method not only accurately predicts ultimate bearing capacity but also captures the nonlinear characteristics of high-dimensional data. As an offline model alternative to finite element methods, the proposed approach holds promise for efficient and real-time prediction of the mechanical behavior of shallow foundations under loading.

Machine learning (ML) has garnered significant attention within the engineering domain. However, engineers without formal ML education or programming expertise may encounter difficulties when attempting to integrate ML into their work processes. This study aims to address this challenge by offering a tutorial that guides readers through the construction of ML models using Python. We introduce three simple datasets and illustrate how to preprocess the data for regression, classification, and clustering tasks. Subsequently, we navigate readers through the model development process utilizing well-established libraries such as NumPy, pandas, scikit-learn, and matplotlib. Each step, including data preparation, model training, validation, and result visualization, is covered with detailed explanations. Furthermore, we explore explainability techniques to help engineers understand the underlying behavior of their models. By the end of this tutorial, readers will have hands-on experience with three fundamental ML tasks and understand how to evaluate and explain the developed models to make engineering projects efficient and transparent.

Soil permeability is a critical parameter that dictates the movement of water through soil, and it impacts processes such as seepage, erosion, slope stability, foundation design, groundwater contamination, and various engineering applications. This study investigates the permeability of soil amended with waste foundry sand (WFS) at a replacement level of 10%. Permeability measurements are conducted for three distinct relative densities, spanning from 65% to 85%. The dataset compiled from these measurements is employed to develop ensemble artificial intelligence (AI) models. Specifically, four regressor AI models are considered: Nearest Neighbor (NNR), Decision Tree (DTR), Random Forest (RFR) and Support Vector Machine (SVR). These models are enhanced with four distinct base learners: Gradient Boosting (GB), Stacking Regressor (SR), AdaBoost Regressor (ADR), and XGBoost (XGB). The input parameters include fraction of base sand (BS), fraction of waste foundry sand (WFS), relative density (RD), duration of flow (T), quantity of flow (Q) and permeability (k), totalling 165 data points. Through comparative analysis, the Gradient Boost with Decision Tree (GB-DTR) model is found to be best-performed model, with R² = 0.9919. Sensitivity analysis reveals that Q is the most influential input parameter in predicting soil permeability.

The utilization of waste residues for soil amelioration is becoming increasingly popular in the construction industry due to its potential for effective waste management and resource utilization. This practice is of utmost importance for the sustainable development of nations, as it offers both environmental protection and economic benefits. In this study, we investigate the sustainable incorporation of Design of Experiment (DOE) to optimize the use of binary additives for enhancing expansive soil. The selected binary additives for this study are calcium carbide residue (CCR) and palm oil fuel residue (POFR). A total of twenty different mix designs were prepared using various combinations of CCR, POFR, water, and soil, following the Scheffe’s DOE strategy. To evaluate the performance and effectiveness of the additives, mechanical testing, including durability and unconfined compressive strength tests, was conducted. The results showed peak values of 58% for durability and 735 kN/m² for unconfined compressive strength (UCS). Additionally, the analysis of variance and student t-test, which are standard techniques for assessing the goodness of fit, were applied to statistically analyse the mathematical models and validate their adequacy and validity. Microstructural experiments, involving scanning electron microscopy (SEM) and Fourier transform infrared spectroscopy (FTIR), were performed on the natural soil and soil treated with the optimal level of additives. The SEM analysis confirmed the formation of new compounds resulting from the incorporation of CCR-POFR mixtures, while the FTIR analysis validated the presence of different molecular functional groups in the treated soil.

The accurate prediction of the fundamental period of steel-braced reinforced concrete (RC) buildings is crucial for optimizing seismic design and ensuring structural safety. Traditionally, empirical formulas provided by building codes such as Eurocode 8 and ASCE 7–22 primarily rely on building height to estimate the fundamental period. However, these height-based models often overlook the significant influence of other structural parameters, such as bracing configurations, bracing lengths, and material properties. This study addresses these limitations by offering a comprehensive evaluation of the factors affecting the fundamental period of steel-braced RC buildings, using advanced computational techniques for more precise and interpretable predictions. A dataset comprising 17,280 building models with varied structural configurations was generated using computational simulations. Key parameters, including total building height, bracing type, bracing length, and building dimensions, were systematically varied. The study utilized machine learning techniques and employed SHapley Additive exPlanations (SHAP) and Individual Conditional Expectation (ICE) plots as post-hoc interpretability tools to analyze the contributions of structural parameters. Results show that total building height remains the dominant factor, contributing approximately 45% to the predicted fundamental period, while bracing length and bracing type significantly influence the period, reducing it by up to 20%. The inclusion of these parameters improves prediction accuracy and reveals limitations in existing height-based formulas. The study concludes that height alone is insufficient for accurate prediction of the fundamental period in steel-braced RC buildings. Incorporating bracing systems and other structural factors is essential for more reliable seismic design. These findings contribute to the development of more resilient building codes and enhanced seismic performance.

Piano Key Weir (PKW) is an advanced hydraulic structure that enhances water discharge efficiency and flood control through its innovative design, which allows for higher flow rates at lower upstream levels. Accurate discharge prediction is crucial for PKW performance within various water management systems. This study assesses the efficacy of Artificial-Neural-Network (ANN) and Gene-Expression-Programming (GEP) models in improving discharge prediction for symmetrical PKWs. A comprehensive dataset comprising 476 experimental records from previously published studies was utilized, considering a range of geometric and fluid parameters (PKW key widths, PKW height, and upstream head). In the training stage, the ANN model demonstrated a superior determination coefficient (R²) of 0.9997 alongside a lower Mean Absolute Percentage Error (MAPE) of 0.74%, whereas the GEP model yielded an R² of 0.9971 and a MAPE of 2.36%. In the subsequent testing stage, both models displayed a high degree of accuracy in comparison to the experimental data, attaining an R² value of 0.9376. Furthermore, SHapley-Additive-exPlanations and Partial-Dependence-Plot analyses were incorporated, revealing that the upstream head exerted the greatest influence on the discharge prediction, followed by PKW height and PKW key width. Therefore, these models are recommended as reliable, robust, and efficient tools for forecasting the discharge of symmetrical PKWs. Additionally, the mathematical expressions and associated script codes developed in this study are made accessible, thus providing hydraulic engineers and researchers with the means to perform rapid and accurate discharge predictions.

This paper presents an intelligent lithology identification method that utilizes the feature fusion of single polarized and orthogonal polarized rock images. The traditional thin section identification method heavily relies on manual expertise, leading to subjective results and requiring significant time and labor. To overcome these limitations, we establish a microscopic feature fusion model using a convolutional neural network (CNN). This model leverages the complementarity information from single polarized and orthogonal polarized features. By extracting features from microscopic rock images using convolutional kernels and integrating multi-feature information at both the input and feature levels, the proposed method enhances the classification accuracy of the model, providing a more efficient and objective solution for lithology identification. To evaluate the identification performance, several metrics including accuracy (Acc), precision (P), recall (R), F1-score, and a confusion matrix are employed. The results demonstrate that the fusion model achieved a maximum accuracy of 98.66% on the testing set, representing a 4.91% improvement over using single polarized images alone and a 1.55% improvement over orthogonal polarized images alone. The integration of advanced deep learning models with microscopic image analysis techniques enables researchers and non-geologists to automate the identification and classification of extensive rock sample datasets efficiently. Moreover, the proposed method proves particularly useful in cases with complex mineral compositions and similar structures, as it provides more reliable and accurate analytical results.

The demand for construction materials in Pakistan has experienced a significant increase, particularly due to the China-Pakistan Economic Corridor (CPEC) project, which necessitates substantial amounts of resilient resources for infrastructure development. Parameters of rock strength, including uniaxial compressive strength (UCS), Young’s modulus (E), and Poisson’s ratio (ν), are critical attributes of rock materials vital for applications such as rock slope stability assessment, tunnel construction, and foundation design. Conventionally, the measurement of UCS, E, and ν in laboratory settings resource-intensive, requiring considerable time and financial investment. This study proposes to provide a comprehensive assessment framework using an adaptive boosting machine (AdaBoost), extreme gradient boosting machine (XGBoost), and category gradient boosting machine (CatBoost), to indirectly estimate UCS, E, and ν through streamlined mineralogical analyses. The performance of the boosting trees was analyzed using Taylor diagrams and a suite of five regression metrics: coefficient of determination (R²), root mean square error (RMSE), mean absolute error (MAE), variance accounted for (VAF), and the A-20 index. The results indicate that the proposed boosting trees robust predictive capabilities for the constructed database. Notably, AdaBoost demonstrated the highest efficacy in predicting the strength of carbonate rock, achieving R² values of 0.98, 0.99, and 0.97, with the lowest RMSE values of 0.3164, 0.63, and 0.18, for UCS, E, and ν, respectively. Moreover, variable importance analysis highlighted that the presence of micrite and calcite has a significant impact on predicting UCS, E, and ν of carbonate rock. Furthermore, the AdaBoost model was validated using an independent dataset, which corroborated its predictive reliability. In conclusion, the proposed models present a highly effective methodology for the indirect prediction of essential mechanical properties of carbonate rocks, offering substantial time and cost efficiencies compared to traditional laboratory techniques.

Geopolymer concrete is acknowledged as a sustainable alternative to conventional Portland cement concrete owing to its ability to reduce carbon emissions and reutilize industrial by-products. This paper reviews the application of Artificial Intelligence-based Life Cycle Analysis (LCA) techniques in the sustainability assessment of geopolymer concrete. The assessment covers the entire life cycle of geopolymer concrete, spanning from the extraction of raw materials to its ultimate disposal, with a particular focus on its environmental, economic, and social impacts. The incorporation of AI techniques into the LCA process offers notable advantages, such as the efficient management of large datasets, enhancement of data quality, prediction of environmental impacts, and facilitation of informed decision-making. Key sustainability metrics to be considered include environmental impacts such as carbon footprint and energy consumption, economic factors like cost-effectiveness, as well as social implications. The amalgamation of AI within the LCA framework provides a comprehensive and efficient approach to evaluating the sustainability of geopolymer concrete, thereby facilitating its application in sustainable construction practices.