Jul 2026· Journal of Science and Transport Technology· 0 citations· 31 references
TL;DR
This study applies several machine learning models to predict the Marshall stability of Stone Mastic Asphalt mixtures incorporating steel slag as a partial replacement for conventional coarse aggregates, with CatBoost providing the best performance.
Abstract
This study applies several machine learning models, including Gradient Boosting Regressor (GBR), CatBoost (CB), Support Vector Machine (SVM), Random Forest (RF), and AdaBoost (AB), to predict the Marshall stability of Stone Mastic Asphalt (SMA) mixtures incorporating steel slag as a partial replacement for conventional coarse aggregates. The dataset consists of 144 Marshall test samples, with input variables including specific gravity, penetration, flash point temperature, softening point temperature, bitumen content, cellulose fiber content, and steel slag content, while Marshall stability is considered as the output variable. The results indicate that all models achieved high predictive accuracy, with CatBoost providing the best performance, achieving R² = 0.936, RMSE = 0.227, and MAE = 0.323 on the validation dataset. Additional analyses, including residual analysis and SHAP-based interpretation, suggest the robustness and interpretability of the model within the investigated dataset of the model. These findings highlight the strong potential of CatBoost for accurately predicting the mechanical performance of SMA mixtures and support decision-making in mixture design in pavement engineering.
An interpretable machine-learning framework for predicting the splitting strength of asphalt concrete and supporting data-driven mixture design and a GUI platform integrating prediction and SHAP-based explanation was developed to improve the accessibility and practical applicability of the proposed framework.
J. Xing, Xiao Tan, Dongzhan Jin et al.· 0 citations
This work addresses the simultaneous prediction of Marshall Stability (MS) and Indirect Tensile Strength (ITS) by integrating machine learning models with multi-objective optimization for the preliminary design of asphalt concrete. Based on 389 experimental samples, 15 variables were selected to describe asphalt properties, aggregate gradation, volumetric parameters and fiber characteristics, and four dual-output prediction models were developed. The models were evaluated using 50 Monte Carlo splits. TabICLv2 performed slightly better for MS prediction, with an RMSE of 1.49 ± 0.22 kN and an R2 of 0.85 ± 0.04, whereas TabPFN showed a slight advantage for ITS prediction, achieving an RMSE of 0.23 ± 0.08 MPa and an R2 of 0.91 ± 0.06. Furthermore, Pareto filtering identified nine non-dominated mixtures, and TOPSIS ranking selected the highest-ranked equal-weight compromise mixture, with MS = 15.23 kN and ITS = 3.90 MPa. The results indicate that mineral fibers are more suitable for improving the balanced performance of MS and ITS, carbon fibers are more favorable for improving MS, and plastic fibers are more effective in improving ITS. Finally, a Streamlit-based graphical user interface was developed to enable real-time prediction and MS–ITS trade-off visualization, providing a reference for preliminary mix design of asphalt concrete.
J. Xing, Xiao Tan, Mu Guo et al.· Materials· 0 citations
This study develops an integrated machine learning-experimental framework to predict the compressive strength (CS) of concrete incorporating ternary industrial wastes glass powder, marble powder, and iron ore slag. For this purpose, a dataset comprising 366 mix ratios and corresponding CS values was compiled from various sources for analysis. Advanced machine learning (ML) algorithms, including extreme gradient boosting (XGB), gradient boosting, and random forest (RF), were employed alongside hybrid techniques such as XGB-GBR and XGB-RF to evaluate the influence of these materials on strength. Based on the outcomes of the analysis, the hybrid XGB-GBR model demonstrates the highest balanced performance for both training (R2 = 0.911) and testing (R2 = 0.869) data sets. For validating the ML modeling and developing an interactive graphical user interface (GUI), experimental evaluation of CS and scanning electron microscopy was conducted. Additionally, feature importance modeling and optimization identified curing age and coarse aggregate as the most influential factors that would impact the model prediction. The contribution of this research lies in the combined modeling and experimental evaluation of a ternary waste concrete system, along with the development of a GUI. This deployable GUI will enhance the industrial applicability of ML-based concrete optimization by reducing material costs, minimizing trial batching, and supporting sustainable mix design practices.
Md. Samsuzzaman Sobuz, Md. Kawsarul Islam Kabbo, Abdullah Alzlfawi et al.· Scientific Reports· 0 citations
Pavement mix design for steel slag relies largely on empirical Marshall tests requiring numerous specimens and lengthy cycles. To address this, an inverse mix design (IMD) framework combining machine-learning forward prediction with grey wolf optimization (GWO) was developed. A dataset of 300 samples with 13 input features and 2 output indicators was compiled. Three algorithms—XGBoost, CatBoost, and random forest (RF)—were compared, and model interpretability was analyzed using SHAP and ALE. CatBoost achieved the best overall performance. SHAP identified steel slag f-CaO content and replacement ratio as the dominant factors governing moisture susceptibility. GWO search errors for all three design scenarios were below 0.24%. Laboratory validation showed a mean deviation of 1.02% between target and measured values, confirming the method’s feasibility. The method also supports sustainable pavement engineering by facilitating higher steel slag utilization, contributing to CO2 reduction and natural aggregate conservation.
The rapid rate of urbanization and industrialization has driven the excessive use of natural resources like river sand and gravel, raising significant sustainability concerns. Waste foundry sand (WFS), a discarded by-product of ferrous and nonferrous metal casting industries, offers a promising substitute for natural sand in concrete. This study focuses on predicting the compressive strength (CS) of WFS-infused concrete by analyzing the impact of various factors, such as cement content, WFS proportion, supplementary cementitious materials (SCMs), water, aggregate composition, and superplasticizer (SP) usage. A data set comprising 401 mix ratios and their corresponding strengths was developed using systematic literature review approach and analyzed using advanced machine-learning (ML) models, including extreme gradient boosting (XGB), categorial boosting (CatB), light gradient boosting, gradient boosting, decision tree,
k
-nearest neighbor, adaptive boosting, bagging regressor, and random forest. The data set was divided into training and testing subsets, and statistical evaluations were performed to determine correlations between input parameters and strength. Among the models, XGB and CatB demonstrated the highest accuracy (
R
2
=
0.98
and 0.97 for training data;
R
2
=
0.83
and 0.86 for testing data, respectively). Shapley additive explanations (SHAP) and partial dependence plot (PDP) analysis revealed that water content and curing age significantly enhanced compressive strength. Furthermore, the developed graphical user interface will help to practically estimate the compressive strength of WFS concrete without any experimental trials.
M. H. R. Sobuz, Md. Kawsarul Islam Kabbo, Abdullah Alzlfawi et al.· Journal of Structural Design...· 0 citations
Accurate and reliable prediction of surface roughness (Ra) is essential for intelligent machining of hardened tool steels under limited-data conditions. This study investigates the influence of cutting and geometric parameters on Ra during the external turning of SKD61 steel using a Taguchi L27 experimental design. The investigated variables included cutting speed, feed rate (f), depth of cut, tool nose radius (r), and workpiece diameter. A comparative machine learning framework was developed to evaluate four prediction models, namely artificial neural network, extreme learning machine (ELM), support vector regression, and random forest regression, with polynomial regression serving as the benchmark. Model performance was first evaluated using repeated 5-fold cross-validation and subsequently assessed using an independent external dataset comprising previously unseen intermediate parameter combinations within the investigated parameter domain. Taguchi analysis and ANOVA identified f and r as the dominant factors affecting Ra. Among the investigated models, ELM achieved the highest prediction accuracy, yielding R2 = 0.9979 ± 0.0006 and MAE = 0.0397 ± 0.0065 µm during repeated validation, together with R2 = 0.9371 and MAE = 0.0845 µm on the independent external dataset. These results support the potential effectiveness of the proposed robustness-oriented evaluation framework for ML-assisted Ra prediction under limited-data machining conditions.
Huynh Thanh Phong, Nguyen Van Thanh Tien, Huynh Thanh Thuong· Engineering Research Express· 0 citations