Journal of Science and Transport Technology
https://jstt.vn/index.php/en
<p><img class="img-responsive" src="https://jstt.vn/public/journals/1/jstt_scopus.png" alt="JSTT has been accepted in Scopus" /></p> <p>Journal of Science and Transport Technology (JSTT) (E-ISSN: <a href="https://portal.issn.org/resource/ISSN/2734-9950">2734-9950</a>) under the publisher of <a href="https://utt.edu.vn/">University of Transport Technology (UTT)</a> has been granted permission by the Ministry of Information and Communication, Vietnam, under Document No. 399/GP-BTTTT dated June 29, 2021, to publish issues in English. JSTT is indexed in <a href="https://www.scopus.com/sourceid/21101274771?origin=resultslist">SCOPUS</a> and <a href="https://scholar.google.com/citations?hl=vi&user=7PS1tesAAAAJ&view_op=list_works&sortby=pubdate">Google Scholar</a>. All published papers are assigned a <a href="https://www.doi.org/">DOI</a> and are registered with <a href="https://www.crossref.org/">Crossref</a>. To ensure academic integrity, each submission is thoroughly checked for similarity using the <a href="https://www.ithenticate.com/">iThenticate</a> tool to prevent plagiarism.</p> <p>JSTT is dedicated to continuously enhancing the quality of its published articles and online editorial system to meet international standards. It serves as a prestigious platform for local and international scientists to exchange and publish new research findings, supporting scientific advancements and industry applications. In its pursuit to solidify its international standing, the Journal is actively seeking contributions from domestic and international scientists.</p> <p>JSTT is an international peer-reviewed multidisciplinary journal dedicated to the advancement of scientific knowledge, technological innovation, and engineering applications relevant to transport systems, infrastructure development, and allied fields of science and technology. The journal provides a platform for researchers, academicians, practitioners, policymakers, and industry professionals to disseminate high-quality original research, reviews, technical communications, and case studies that contribute to scientific understanding and sustainable technological development.</p> <p>JSTT promotes interdisciplinary research that integrates fundamental sciences, engineering disciplines, environmental considerations, digital technologies, and management approaches to address contemporary challenges associated with transportation, infrastructure systems, urban development, and societal progress. The journal encourages both theoretical and applied studies that advance innovation, sustainability, resilience, safety, and operational efficiency at local, regional, and global scales.</p> <p>JSTT publishes original research articles, review papers, technical notes, and case studies in multidisciplinary areas of science, engineering, technology, and management related to transport systems, infrastructure development, and associated scientific applications.</p> <p>The journal welcomes contributions in, but is not limited to, the following fields:</p> <p> Transportation and Traffic Engineering<br /> Civil and Infrastructure Engineering<br /> Geotechnical and Geological Engineering<br /> Construction Materials and Technologies<br /> Mechanical and Automotive Engineering<br /> Electrical, Electronics and Communication Engineering<br /> Computer Science and Information Technology<br /> Environmental and Earth Sciences<br /> Coastal, Hydraulic and Water Resources Engineering<br /> Architecture and Urban Planning<br /> Economics, Management and Policy Studies</p> <ul> <li><a href="https://jstt.vn/index.php/en/about#aim-and-scope"><strong>Aim and scope</strong></a></li> <li><a href="https://jstt.vn/index.php/en/about#peer_review_process"><strong>Peer Review Process</strong></a></li> <li><strong><a href="https://jstt.vn/index.php/en/about#public_frequency">Publication Frequency</a><br /></strong></li> <li><a href="https://jstt.vn/index.php/en/about#article_processing_charge"><strong>Article Processing Charge</strong></a></li> <li><a href="https://jstt.vn/index.php/en/about#licence"><strong>License</strong></a></li> <li><a href="https://jstt.vn/index.php/en/publication_ethics"><strong>Publication Ethics and Malpractice Statement</strong></a></li> <li><a href="https://jstt.vn/index.php/en/guide-for-authors"><strong>Guide for authors</strong></a></li> <li><a href="https://jstt.vn/index.php/en/about#journal-policies"><strong>About the Journal</strong></a></li> </ul>University of Transport Technologyen-USJournal of Science and Transport Technology2734-9950A hybrid linear regression and CatBoost model for high-performance nonlinear static analysis of functionally graded plates
https://jstt.vn/index.php/en/article/view/1286
<p>Nonlinear static analysis of functionally graded material (FGM) plates using high-fidelity numerical methods such as Isogeometric Analysis (IGA) is computationally demanding because it requires iterative solution procedures to trace the equilibrium path. Recently, deep learning surrogates (e.g., Long Short-Term Memory, LSTM) have been explored to reduce repeated simulations; however, they can still require nontrivial training effort and careful model tuning. This study proposes a high-performance hybrid surrogate model combining Linear Regression (LR) and CatBoost to predict the nonlinear load-deflection behavior of FGM plates. The equilibrium-path response is formulated as a sequence regression task along pseudo-time load steps and decomposed into two stages: LR captures the global linear trend, while CatBoost learns the nonlinear residual component not represented by LR. The proposed approach is trained and validated using IGA-generated datasets. It is evaluated under different data-availability settings (30%, 50%, and 80% of the equilibrium-path sequence used for training and testing) and assessed in terms of prediction accuracy, hyperparameter sensitivity, and end-to-end computational cost (training + prediction). In our implementation, the combined LR-CatBoost training and prediction time is approximately 0.25 s for the baseline case, excluding the IGA computation required to generate the initially available response segment. Prediction accuracy is high, with mean absolute percentage errors on the order of 10<sup>-2</sup> to 10<sup>-1</sup>% depending on the case. These findings indicate that the proposed LR-CatBoost strategy provides a robust and ultra-fast surrogate for near-real-time nonlinear structural analysis of FGM plates.</p>Do Thi Thanh DieuSon Thai
Copyright (c) 2026 Journal of Science and Transport Technology
2026-10-022026-10-0211710.58845/jstt.utt.2026.en.6.5.1-17Landslide susceptibility mapping using deep learning in Dak Lak, Vietnam
https://jstt.vn/index.php/en/article/view/1563
<p>Landslides cause significant damage to transportation infrastructure, settlements, and agricultural land in the Central Highlands of Vietnam, particularly in the former Dak Lak province. Therefore, identifying areas with the potential for landslide occurrence is essential to support land-use planning and the development of landslide mitigation measures. This study proposes an explainable deep-learning framework for landslide susceptibility assessment using four models: Convolutional Neural Network (CNN), Deep Residual Dense Network (DRDN), Artificial Neural Network (ANN), and Kolmogorov–Arnold Network (KAN). An inventory comprising 235 landslide polygons was analyzed together with 14 conditioning factors, including aspect, slope, curvature, elevation, geology, land use, soil type, rainfall, distance to faults, distance to roads, distance to rivers, the normalized difference vegetation index (NDVI), the topographic wetness index (TWI), and flow accumulation. The results showed that CNN achieved the best performance, with an area under the curve (AUC) = 0.913, a mean absolute error (MAE) = 0.197, and a root mean square error (RMSE) = 0.444 for the training dataset, and AUC = 0.888, MAE = 0.251, and RMSE = 0.501 for the validation dataset. Model interpretation showed that landslide susceptibility was notably influenced by slope, mean annual rainfall, geological factors, and distance to roads. The effects of these factors were nonlinear and varied across different value ranges, while their response patterns also differed among individual locations. A spatial accuracy assessment of the susceptibility map showed that the high and very high susceptibility classes occupied only 18.30% of the study area but contained 94.88% of the observed landslide pixels. In particular, the very high susceptibility class occupied 7.51% of the study area, contained 68.18% of the landslide pixels, and recorded the highest frequency ratio of 9.079. The results demonstrate that explainable deep-learning models can not only support the identification of areas with potential for landslide occurrence but also clarify the roles and response patterns of the conditioning factors, thereby providing a basis for land-use planning and the development of landslide mitigation measures in the study area.</p>Dam Duc NguyenHoang Ngoc LongHiep Van LeTran Van PhongVuong Hong NhatTran Xuan BienPham The TrinhNguyen Hai Ha
Copyright (c) 2026 Journal of Science and Transport Technology
2026-11-102026-11-1030433010.58845/jstt.utt.2026.en.6.5.304-330The effect of tuned liquid damper placement on the seismic response of a high-rise building: A reduced-order modelling study
https://jstt.vn/index.php/en/article/view/1443
<p>High-fidelity fluid–structure interaction (FSI) analysis is widely recognized as one of the most accurate approaches for evaluating the seismic performance of tuned liquid dampers (TLDs). However, coupling a three-dimensional FSI model with a detailed finite-element model of a high-rise building requires substantial computational resources, limiting its application in routine engineering analysis and preliminary design. This study develops and validates a computationally efficient reduced-order structural–FSI framework that preserves the dominant structural dynamic characteristics while explicitly modeling three-dimensional liquid sloshing. A 30-story reinforced-concrete building is first modeled in ETABS to extract its modal properties. The detailed structural model is subsequently transformed into an equivalent lumped mass–spring–damper system implemented in ANSYS and coupled with three-dimensional TLD fluid domains using FLUID80 elements. The reduced-order structural model reproduces the fundamental natural frequency of the full three-dimensional model with an error of less than 0.2%, demonstrating excellent agreement in the dominant dynamic response. Nonlinear time-history analyses are performed under the 1940 El Centro north–south earthquake record to investigate two practical TLD layouts: six tanks installed at the roof and a distributed configuration with three tanks at the roof and three at mid-height. Both configurations reduce global structural displacement and generally mitigate interstory drift and story shear, although local increases in drift or shear can occur at some floors. The roof-mounted configuration decreases the peak roof displacement from 0.389 m to 0.352 m (approximately 9.5%) and reduces the maximum base shear from 5.52 x 10<sup>7</sup> N to 5.22 x 10<sup>7 </sup>N. Although the distributed configuration provides smaller global reductions, it offers improved response mitigation at intermediate stories through enhanced interaction with higher vibration modes. The proposed framework provides a mechanically consistent and computationally practical methodology for system-level seismic assessment and preliminary design of high-rise buildings incorporating tuned liquid dampers.</p>Thuong Anh NguyenAn Le HoangSang Pham NgocHoa Thi Trinh
Copyright (c) 2026 Journal of Science and Transport Technology
2026-11-102026-11-1027630310.58845/jstt.utt.2026.en.6.5.276-303Experimental study on the performance of a slotted wells turbine in an oscillating water column device
https://jstt.vn/index.php/en/article/view/1424
<p>Ocean wave energy has garnered significant attention as a sustainable power source, with the Oscillating Water Column (OWC) device emerging as a highly effective technology for wave-to-electricity conversion. Within this system, the Wells turbine serves as a critical component for energy extraction. While current literature predominantly relies on Computational Fluid Dynamics (CFD) simulations to analyze the performance of these turbines, this study employs a rigorous experimental approach to evaluate their dynamics. The experimental results demonstrate high reliability when compared with CFD simulation data. Furthermore, this research evaluates the performance of a slotted Wells turbine design. The comparative analysis reveals that the slotted Wells turbine achieves superior efficiency compared to the conventional Wells turbine design. These findings offer a comprehensive empirical foundation and provide critical insights for the future optimization of wave energy converters. Ultimately, this study serves as an essential precursor to subsequent developments aimed at enhancing the operational viability of OWC systems in practical, real-world applications.</p>Thuc Tri DangHuy Hoang TruongVay Siu LoThien Tich Truong
Copyright (c) 2026 Journal of Science and Transport Technology
2026-10-142026-10-14516210.58845/jstt.utt.2026.en.6.5.51-62Advancing construction safety analysis: A comparative study of SBERT-KMeans and LDA in extracting highway work-zone accident topics
https://jstt.vn/index.php/en/article/view/1399
<p>Highway work zones are widely recognized as some of the most hazardous occupational environments due to the simultaneous presence of construction activities and continuous traffic flow. The U.S. Occupational Safety and Health Administration (OSHA) accident reports are incredibly valuable for tracking safety risks and incident trends. Many studies have used latent Dirichlet allocation (LDA) to analyze these documents, but LDA does not usually capture the deeper semantic context and cannot handle the messy and diverse descriptions of accidents. This study explores an alternative approach to this issue by applying SBERT-KMeans topic modeling to analyze 1,051 OSHA reports from January 2015 to February 2022. The performance of SBERT-KMeans is first compared with a traditional Latent Dirichlet allocation (LDA) model. In addition to the classical measures of topic coherence and topic diversity, which are often optimized for single-word representations, the present study also evaluates the models' ability to preserve syntactic structure and recognize complex technical phrases. The results indicate that SBERT-KMeans provides a more phrase-aware representation of domain-specific concepts and technical expressions. Moreover, the extracted safety-risk-related topics are integrated with structured tabular data to evaluate severity correlations (hospitalization and amputation rates), while a topic semantic similarity network is employed to uncover highly interconnected topic hubs. The comparison suggests that SBERT-KMeans provides phrase-aware semantic representations that complement the unigram-based LDA baseline and can support data-driven safety analysis for highway construction management.</p>Sang NguyenTuyen Le
Copyright (c) 2026 Journal of Science and Transport Technology
2026-11-092026-11-0926027510.58845/jstt.utt.2026.en.6.5.260-275Object-based LULC classification and urban dynamics analysis using machine learning on sentinel-2 imagery: A case study of selected wards in Hanoi, Vietnam
https://jstt.vn/index.php/en/article/view/1385
<p>This study investigates land use/land cover (LULC) classification and urban spatial dynamics in three wards: Thanh Xuan, Khuong Dinh, and Phuong Liet, Hanoi, Vietnam, using object-based image analysis (GEOBIA) integrated with machine learning on Sentinel-2 imagery. Sentinel-2 images acquired in 2015, 2020, and 2025 were preprocessed, segmented using the Simple Non-Iterative Clustering (SNIC) algorithm, and used to extract spectral, textural, and shape features at the object level. Four classifications, algorithms including Random Forest (RF), Support Vector Machine (SVM), Classification and Regression Trees (CART), and Gradient Tree Boosting (GTB) were applied and compared. All models achieved satisfactory accuracy; however, RF outperformed the others and showed the most consistent validation results, with overall accuracy ranging from 82.62% to 87.35% and Kappa coefficients from 0.793 to 0.831. The RF-based results indicate that built-up land is the dominant class and has expanded significantly, increasing from 91.04 ha in 2015 to 100.88 ha in 2025, while water bodies, shrubland, agricultural land, and bare land declined. These findings demonstrate the effectiveness of GEOBIA combined with machine learning for analyzing LULC changes in complex urban environments.</p>Le Thi Minh PhuongNguyen Thanh LenLe Huyen Trang
Copyright (c) 2026 Journal of Science and Transport Technology
2026-11-122026-11-1233134810.58845/jstt.utt.2026.en.6.5.331-348Comparative assessment of EN 1994-1-1, AISC 360 and JGJ 138 for multi-profile composite mega-columns, with a parametric extrapolation towards ultra-high-performance concrete
https://jstt.vn/index.php/en/article/view/1375
<p>Mega-columns that embed several discrete hot-rolled profiles in a single reinforced-concrete section, sometimes called isolated steel reinforced concrete (ISRC), have increasingly been adopted in place of welded steel caissons in super-tall lower storeys, but their design falls in a regulatory gap. The simplified method of EN 1994-1-1 covers only single-profile sections, AISC 360-22 admits multiple encased shapes provided they are interconnected but caps the cylinder strength at 69 MPa, and JGJ 138-2016 is confined to concrete grades up to about C80. This study examines how far those provisions can be extrapolated, using a closed-form code-based assessment of a nine-specimen database that combines the six monotonic Phase-1 columns of Bogdan et al. with three axial multi-profile columns of Zhao et al. A single plastic-stress-distribution engine builds the four-point interaction polygon identically under each code, so the only difference is the concrete treatment. All predictions are computed at the characteristic (measured) material level, with partial safety factors set to unity, so that the comparison isolates the cross-sectional mechanics and does not represent factored design capacities. EN 1994-1-1 and AISC 360-22 both apply the 0.85 effective-stress block and coincide, at a mean test-to-prediction ratio λ = 1.12 (CoV 13.4%) against 1.02 for JGJ 138-2016 (CoV 12.2%), whose near-unity α<sub>1</sub> makes it least conservative. The predictions are generally conservative across the eccentric range and for the Zhao cage, the single exception being E10-2 under JGJ 138-2016 (λ = 0.977), but they overpredict Bogdan's two pure-axial specimens at f’<sub>c</sub> ≈ 55-61 MPa by 11–14% (EN/AISC) and 19–23% (JGJ), reflecting the non-attainment of the full plastic state at peak–the steel still at or below yield and the cover spalling prematurely under full-area loading. In the parametric sweep at e/h = 12.5%, the JGJ margin over the coincident EN/AISC prediction holds within 7–10% up to 180 MPa. Extending these rules to UHPC will require a calibrated effective concrete stress, a minimum steel-fibre dosage, and explicit treatment of eccentricities up to e/h ≈ 15%.</p>Hang Mai-ThiThi Thu Tra Nguyen
Copyright (c) 2026 Journal of Science and Transport Technology
2026-11-092026-11-0923425910.58845/jstt.utt.2026.en.6.5.234-259Stress-dependent SPT-based estimation of small-strain shear modulus for marine sand in central Vietnam and application to high-speed railway foundations
https://jstt.vn/index.php/en/article/view/1371
<p>This study evaluates the influence of effective stress on the small-strain shear modulus (G₀) of coastal sand in central Vietnam using data obtained from downhole seismic and standard penetration tests (SPT). A database consisting of 50 paired SPT–downhole test intervals collected from the investigated boreholes was analyzed to develop a stress-dependent correlation of the form G₀ = A(σ’<sub>v</sub>)<sup>m</sup>N<sup>n</sup>, where G₀ is expressed in MPa and σ’v in kPa. Regression analysis indicates that the contribution of SPT resistance (n ≈ 0.40) is substantially greater than that of effective stress (m ≈ 0.20), suggesting that soil shear stiffness is governed primarily by penetration resistance. The weak stress dependency explains why conventional G₀–N correlations provide reasonable predictions for the investigated deposits. The proposed formulation offers a practical framework for preliminary stiffness estimation and dynamic behavior assessment of coastal sandy soils under field conditions. To demonstrate its engineering applicability, the estimated G₀ profiles were incorporated into a layered elastic railway foundation model based on the analytical approach proposed by Sheng et al. (1999). Representative loading conditions corresponding to the future North–South High-Speed Railway in Vietnam, including passenger and freight axle loads, were considered. Parametric analyses were conducted to evaluate the effects of train speed, and axle load on the vertical displacement response of the foundation system. The results indicate that the proposed correlation can provide realistic stiffness profiles for preliminary evaluation of railway foundation performance and can serve as a practical tool for geotechnical assessment in regions where direct geophysical measurements are unavailable.</p>Dang Hong LamPhi Hong ThinhTran Khanh HoanNguyen Chau LanCao Van DoanDo Duc Phi
Copyright (c) 2026 Journal of Science and Transport Technology
2026-11-142026-11-1434936310.58845/jstt.utt.2026.en.6.5.349-363Interpretable hybrid data-driven modeling for predicting the compressive strength of ready-mixed concrete
https://jstt.vn/index.php/en/article/view/1347
<p>This research uses a large dataset of ready‑mixed concrete batches collected directly from projects in Vietnam to develop a data-driven modeling framework for compressive‑strength prediction. The dataset comprises 1,157 industrial samples and includes ten input variables describing mixture proportions, aggregate gradation, admixture dosage, and fresh concrete workability. Seven models were evaluated, namely Random Forest (RF), Artificial Neural Network (ANN), Gradient Boosted Regression Trees (GBRT), Extreme Gradient Boosting (XGBoost), Model Tree (M5P-like), Support Vector Regression (SVR), and Gaussian Process Regression (GPR). All models were trained using an 80/20 train–test split, with hyperparameters optimized through Bayesian optimization to ensure robust performance. Among the evaluated models, GPR demonstrated the best overall generalization capability, achieving a coefficient of determination (R²) of 0.9399, a root mean square error (RMSE) of 3.042 MPa, and a mean absolute error (MAE) of 1.883 MPa. Extreme Gradient Boosting (XGBoost) achieved the lowest prediction error in terms of mean absolute error (MAE = 1.756 MPa) and mean absolute percentage error (MAPE = 4.57%), indicating strong predictive accuracy. To enhance model transparency, permutation-based global feature importance analysis and local surrogate interpretation techniques were applied to identify the key factors influencing model predictions. The results reveal that the aggregate-to-cement ratio, cement content, admixture dosage, and water-to-cement ratio are the most influential variables governing compressive strength. Additionally, local prediction behavior is notably affected by curing age, coarse aggregate content, and slump under specific conditions. Overall, the findings confirm that interpretable machine learning models, particularly Gaussian Process Regression and Extreme Gradient Boosting, provide reliable and practical tools for strength prediction, mixture optimization, and production-oriented quality control in real-world construction applications in Vietnam.</p>Nguyen Hoang LongMay Huu NguyenAnh Nguyen HuuDung Quang Vu
Copyright (c) 2026 Journal of Science and Transport Technology
2026-10-222026-10-2215017910.58845/jstt.utt.2026.en.6.5.150-179Characterization of anisotropy in sheet metals using digital image correlation
https://jstt.vn/index.php/en/article/view/1327
<p>This study develops and applies an in-house two-dimensional digital image correlation (2D-DIC) system to investigate the anisotropic behavior of sheet metals under uniaxial tensile loading. Two materials, SUS304 stainless steel and DP1200 dual-phase steel, were examined. Tensile specimens were prepared along three orientations with respect to the rolling direction: rolling direction (RD), diagonal direction (DD), and transverse direction (TD), and the tests were performed in accordance with ISO 6892-1. The DIC method provided full-field displacement and strain measurements on the specimen surface, enabling the determination of stress–strain curves, Lankford coefficients, normalized yield stresses, and yield surface characteristics. The results show that SUS304 exhibits high ductility, strong strain hardening capacity, and more pronounced planar anisotropy than DP1200. In contrast, DP1200 shows higher tensile strength but lower elongation, and its Lankford coefficients and normalized yield stresses are closer to isotropic behavior. Comparison of the yield criteria indicates that Hill48 represents the anisotropic response more accurately than the von Mises criterion. Hill48-R agrees better with the experimental Lankford coefficients, whereas Hill48-S more accurately describes the normalized yield stresses.</p>Dinh Van TranQuang Huy DuongDuc Anh HaVan Duy DinhQuoc Tuan Pham
Copyright (c) 2026 Journal of Science and Transport Technology
2026-10-162026-10-1611212610.58845/jstt.utt.2026.en.6.5.112-126Interpretable hybrid machine learning frameworks for predicting the axial load capacity of concrete filled double skin steel tube columns
https://jstt.vn/index.php/en/article/view/713
<p>Predicting the axial load capacity of concrete-filled double-skin steel tube (CFDST) columns is difficult because of the complex mechanical interactions between the concrete core and the dual steel tubes. Although machine learning (ML) methods provide advantages in prediction, current models frequently lack sufficient hyperparameter optimization and mechanistic interpretability. To resolve these limitations, novel hybrid ML frameworks are proposed integrating eXtreme Gradient Boosting (XGBoost) with advanced metaheuristic algorithms, specifically the Bald Eagle Search (BES) and Aquila Optimizer (AO). A database containing 153 experimental samples defined by eight geometric and material parameters was used to develop and validate the models. Systematic optimization comparisons demonstrated that the XGB-BES framework provides high generalization and accuracy, producing a coefficient of determination (R<sup>2</sup>) of 0.998, a Root Mean Square Error (RMSE) of 62.054 kN, and an A20 index of 0.957 when tested on validation data. Furthermore, the inherent opacity of complex ML models was addressed using post-hoc interpretability methods such as SHAP (Shapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations). Analyses of global and local feature importance demonstrated that the diameter and thickness of the outer tube most strongly affect the structural capacity, which connects the computational predictions to fundamental structural mechanics principles. These findings indicate that the developed XGB-BES model functions as an accurate and mechanically transparent instrument for the evaluation and design of CFDST columns.</p>Dang Khoa DoThuy-Anh NguyenTrinh Quoc NgoViet-Hung Phan
Copyright (c) 2026 Journal of Science and Transport Technology
2026-10-192026-10-1912714910.58845/jstt.utt.2026.en.6.5.127-149An ensemble JCHAIDStar framework for improved landslide susceptibility prediction using Bagging and Random Subspace techniques
https://jstt.vn/index.php/en/article/view/1239
<p>This study developed a landslide susceptibility map for the mountainous area of northern Vietnam, located within the administrative boundary of the former Ha Giang Province and now under Tuyen Quang Province, to support the identification of landslide-prone areas and disaster risk reduction. The input data included a landslide inventory of 963 locations and 13 conditioning factors representing topographic, hydrological, geological, climatic, and human-related conditions. The dataset was divided into training and validation subsets using a 70:30 ratio. Four models, including Bagging-based JCHAIDStar (Bag-JCHAIDStar), Random Subspace-based JCHAIDStar (RSS-JCHAIDStar), Deep Neural Network (DNN), and Multilayer Perceptron (MLP), were developed and compared using model performance metrics and Taylor diagram analysis. The results show that Bag-JCHAIDStar achieved the best overall performance, with AUC = 0.988 for the training dataset and AUC = 0.950 for the validation dataset. On the validation dataset, this model also obtained ACC = 88.7%, Kappa = 0.760, and RMSE = 0.291, indicating good classification and generalization capability. The validation of the landslide susceptibility zoning map using the Frequency Ratio method showed that the very high susceptibility class had the highest landslide proportion, accounting for 52.11% of the total landslide pixels and reaching the highest FR value of 1.817. These results suggest that the Bag-JCHAIDStar model can provide useful support for landslide susceptibility assessment, disaster risk reduction, and sustainable land-use planning in mountainous areas.</p>Vu Cao DatTran Hong ThaiPham Thi Thanh NgaPham Quang Dung
Copyright (c) 2026 Journal of Science and Transport Technology
2026-10-152026-10-15638510.58845/jstt.utt.2026.en.6.5.63-85Machine learning based prediction of automotive fuel economy
https://jstt.vn/index.php/en/article/view/1229
<p>Accurate prediction of automotive fuel economy is important for improving vehicle design, reducing operating costs, and supporting energy-efficient transportation systems. Conventional empirical and statistical approaches often fail to capture the complex nonlinear relationships among vehicle technical characteristics. This study presents a comparative evaluation of four state-of-the-art ensemble machine learning algorithms, namely Random Forest (RF), Light Gradient Boosting Machine (LGB), XGBoost (XGB), and CatBoost (CGB), for predicting vehicle fuel economy using the publicly available UCI Auto MPG dataset, comprising 398 vehicle records characterized by seven technical attributes (392 valid samples after preprocessing). The models were optimized using Grid search optimization (GSO) and evaluated through an 80:20 training-testing split combined with five-fold cross-validation. Model performance was assessed using the coefficient of determination (R²), root mean square error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE). Model interpretability was investigated using SHapley Additive exPlanations (SHAP) to quantify the influence of individual vehicle attributes. Among the evaluated algorithms, CGB achieved the best predictive performance with a testing R² of 0.9204, normalized RMSE of 0.0473, normalized MAE of 0.0347, and MAPE of 9.07% on the normalized target scale, indicating promising performance on the held-out test set. SHAP analysis revealed that model year, vehicle weight, horsepower, and engine displacement were the most influential variables affecting fuel economy predictions. The study demonstrates that optimized ensemble learning methods can accurately model complex relationships between vehicle specifications and fuel economy while providing interpretable predictions. Although the findings are based on a benchmark dataset representing vehicles manufactured between 1970 and 1982, the proposed comparative framework provides a transparent comparative framework that can be extended to larger and more contemporary automotive datasets.</p>Sushant Kumar SinghNguyen Van HauTien Hung NguyenHoang-Minh LuuIndra Prakash
Copyright (c) 2026 Journal of Science and Transport Technology
2026-10-292026-10-2919721310.58845/jstt.utt.2026.en.6.5.197-213Enhancing multi-sensor classification robustness under noise via adaptive attention fusion
https://jstt.vn/index.php/en/article/view/1205
<p>Vehicle classification is an important part of Intelligent Transportation Systems (ITS). While vision-based sensors struggle in harsh weather, non-vision sensors like PIR and LiDAR offer robust additional data. However, conventional fixed-weight fusion strategies often perform poorly when a single sensor becomes noisy. To address this, we propose an Attention-based Adaptive Sensor Fusion (AASF) model that can dynamically adjust fusion weights using a lightweight attention mechanism. The dataset includes 3,963 samples labeled for four main vehicle types: Buses, Cars, Motorbikes, and Trucks. According to experimental results, the AASF model achieved the highest observed accuracy of 96.3%, compared with 95.8% for CNN fusion and 94.4% for fixed-weight fusion. Under clean conditions, its performance was statistically comparable to CNN fusion, while its main advantage was observed under increasing sensor noise. When compared to data from individual sensors, the combined features from the suggested model show a distinct cluster structure and improved separability between vehicle classes through t-SNE visualization. The work specifically shows that the adaptive attention mechanism retains greater robustness against Gaussian noise, enabling the model to automatically modify weights to prioritize sensor channels that are less impacted by noise. This result confirms the potential of sensor aggregation in improving the efficiency of intelligent transportation systems (ITS).</p>Nguyen Van TuanChu Tien DatVu Van QuangBui Hai DangNguyen Minh TriNguyen Duc SonVu Toan Thang
Copyright (c) 2026 Journal of Science and Transport Technology
2026-11-162026-11-1636437710.58845/jstt.utt.2026.en.6.5.364-377An efficient hybrid framework for reliability-based size and shape optimization of trusses considering uncertainties
https://jstt.vn/index.php/en/article/view/1197
<p>Reliability-based design optimization (RBDO) of truss structures has become a powerful tool for achieving optimal designs under uncertainties in material properties, cross-sectional areas, and applied loadings. By integrating the sequential optimization and reliability assessment (SORA) into an efficient optimization tool based on the success-history-based adaptive differential evolution (SHADE) algorithm, this paper presents a novel RBDO methodology, which is the so-called SORA-SHADE, for the optimization of shape and size in truss structures under uncertainties of input data. The proposed paradigm is tested on well-known 2D and 3D trusses with and without the reliability analyses for validation. For reliability-based analysis cases, each cross-sectional area is considered as a random variable under uncertainties. The results show that the proposed approach is highly effective and reliable in finding optimal designs that meet high reliability requirements.</p>Anh Hoang NguyenQui Xuan LieuVan Quan TranQuan Minh Lieu
Copyright (c) 2026 Journal of Science and Transport Technology
2026-10-042026-10-04183210.58845/jstt.utt.2026.en.6.5.18-32Evaluation of ABS braking performance for vehicles descending slopes under different road adhesion coefficients
https://jstt.vn/index.php/en/article/view/1078
<p>Braking safety on downhill roads with varying tire–road adhesion is a critical concern in vehicle dynamics. This paper evaluates the effectiveness of the Anti-lock Braking System (ABS) during downhill braking through a combined simulation and experimental approach. A longitudinal vehicle dynamics model is developed, integrating a driver model, vehicle body dynamics, wheel dynamics, and an ABS model employing a Fuzzy Logic controller. Simulation results are compared with experimental braking data to validate the model, assess controller performance, and evaluate ABS behavior on an actual vehicle. The comparison shows that braking performance parameters follow similar variation trends in both simulation and experiment; however, the Fuzzy controller maintains the wheel slip ratio within its optimal range with noticeably higher stability in simulation than in experiment. Using the validated model, braking performance from an initial speed of 60 km/h is investigated across different road adhesion coefficients and slope angles. When the adhesion coefficient decreases from 0.8 to 0.3, braking distance increases by 354.07% at an 8° slope and 725.50% at a 12° slope. Under low-adhesion conditions (φ = 0.3), increasing the slope from 8° to 12° raises braking distance by 121.16%, whereas under high-adhesion conditions (φ = 0.8), the increase is only 16.03%. These findings highlight the necessity of simultaneously accounting for road grade and adhesion coefficient when assessing ABS braking performance on downhill roads.</p>Trung Kien NguyenTrong Hoan Nguyen
Copyright (c) 2026 Journal of Science and Transport Technology
2026-10-242026-10-2418019610.58845/jstt.utt.2026.en.6.5.180-196Foam-based eco-lightweight controlled low-strength material for high-volume utilization of municipal solid waste incineration bottom ash
https://jstt.vn/index.php/en/article/view/1045
<p>The increasing scarcity of natural aggregates, coupled with the growing environmental burden associated with municipal solid waste incineration (MSWI) residues, poses a dual challenge for sustainable urban construction. Previous studies have primarily investigated MSWI bottom ash (IBA) or foam-modified mortars in isolation; however, the combined effects of IBA, fly ash, and pre-formed foam on controlled low-strength material (CLSM) have not yet been systematically evaluated. This study develops an integrated, waste-valorized CLSM system and systematically investigates the fresh-state, hardened, durability, and eco-mechanical performance of fifteen mixtures incorporating 0–40% IBA, 0–20% fly ash, and 0–20% pre-formed foam. Cylindrical specimens were tested in accordance with relevant ASTM standards to evaluate flowability, bleeding, setting time, water absorption, 3–28-day compressive strength, and durability under simulated acid rain and seawater immersion, supported by a normalized multi-criteria assessment (MCA) and eco-mechanical indicators (relative carbon emission [RCE] and relative energy consumption [REC]). The results indicate that increasing IBA content reduced flowability from 170 mm to 130 mm and increased water absorption by up to 49%, whereas foam incorporation significantly enhanced flowability to a maximum of 270 mm, albeit with an associated increase in water absorption to 5.44–5.91%. The 28-day compressive strength decreased from 3.213 MPa in the control mixture to 2.187 MPa at the highest combined waste and foam content; nevertheless, all mixtures exceeded the minimum 1 MPa requirement for CLSM applications. Durability losses ranged from 4.77% to 10.51% under acid rain exposure and from 8.34% to 16.61% under seawater immersion, with high-porosity mixtures being the most adversely affected. Despite these trade-offs, mixtures with moderate waste contents—particularly M5 (10% IBA, 10% fly ash, and 10% foam)—exhibited a well-balanced performance, achieving an MCA score of 0.66, a 28-day compressive strength exceeding 2.5 MPa, a flowability of approximately 210 mm, and 85–88% strength retention under aggressive environments. Overall, the findings indicate that the combined use of IBA, fly ash, and pre-formed foam can produce lightweight CLSM with a favorable balance between engineering performance and strength-normalized environmental efficiency, making it promising for non-structural applications such as coastal and municipal backfilling.</p>Trong-Phuoc HuynhKhoa Dang TruongLai Bao TanTri Ho Minh Le
Copyright (c) 2026 Journal of Science and Transport Technology
2026-10-152026-10-158611110.58845/jstt.utt.2026.en.6.5.86-111Numerical study on the effect of ECC on shear behaviour of steel-reinforced short beams
https://jstt.vn/index.php/en/article/view/898
<p>This study presents a numerical analysis of the shear behaviour of steel-reinforced beams made of Engineered Cementitious Composites (ECC) material. The numerical model was developed and validated using the shear performance data of three reinforced concrete (RC) short beams and one reinforced ECC (R/ECC) short beam reported in the literature. Following the validation, a parametric study was conducted to examine and evaluate the effects of ECC material properties. First, the material used in beams was changed from concrete to ECC with an equivalent nominal compressive strength, and then, by altering the ECC compressive strength, tensile strength, and tensile strain, the effects of these parameters were systematically evaluated. The results showed that replacing conventional concrete with ECC significantly enhanced the ultimate shear capacity, with shear load gains of up to 60% over the RC counterparts. By altering the ECC compressive strength, tensile strength, and tensile strain values, the subsequent parametric study revealed that the ECC tensile strength (f<sub>t</sub>) was the dominant material parameter, leading to substantial increases in the ultimate shear load (up to 112% when f<sub>t</sub> was increased from 1.37 MPa to 5.37 MPa). Meanwhile, the tensile strain capacity played a crucial role in enhancing the overall structural ductility and energy absorption of the R/ECC short beams. The study also found that the ECC compressive strength (f’<sub>cr</sub>) had a marginal effect on the shear performance of R/ECC short beams.</p>Cong-Luyen NguyenNgoc Han NguyenDuc Nhan HoangQuyen Cao Minh
Copyright (c) 2026 Journal of Science and Transport Technology
2026-11-062026-11-0621423310.58845/jstt.utt.2026.en.6.5.214-233Enhancing landslide susceptibility prediction using deep learning through optimal activation and loss function selection
https://jstt.vn/index.php/en/article/view/755
<p>Landslide susceptibility prediction has increasingly shifted from traditional statistical methods toward deep learning approaches capable of modeling complex nonlinear relationships among geo-environmental controlling factors. However, model performance is influenced not only by network architecture but also by the selection of activation and loss functions, which directly affect learning dynamics and generalization capability. This study systematically investigates the influence of activation–loss function combinations within a unified deep learning framework for landslide susceptibility mapping in Sin Ho area, northwestern Vietnam. A multi-source geospatial database integrating topographic, geological, hydrological, land-use, seismic, and rainfall-related factors was used to train and validate multiple model configurations. Model performance was evaluated using a comprehensive set of quantitative indicators, including area under the ROC curve (AUC) and accuracy (ACC). The results indicate that models employing the Softmax activation function consistently achieve superior predictive performance and robustness. Among all tested configurations, the Softmax–MultiLabel (SM–ML) model delivered the most reliable results, attaining a validation AUC of 0.878, and ACC of 83.75, demonstrating strong discrimination ability, stable generalization, and reduced prediction error. The landslide susceptibility map generated using the optimal SM–ML configuration exhibits a clear and progressive increase in landslide density from very low to very high susceptibility classes, confirming meaningful spatial stratification of landslide hazard. These findings highlight that careful selection of activation and loss functions can substantially enhance deep learning performance without increasing model complexity or computational cost, providing a robust and efficient framework for landslide susceptibility assessment and disaster-risk reduction in mountainous regions under changing climatic conditions.</p>Tran Huu AnhTran Van PhongDung Nguyen DuyNguyen Thi Bich Hanh
Copyright (c) 2026 Journal of Science and Transport Technology
2026-10-052026-10-05335010.58845/jstt.utt.2026.en.6.5.33-50