4th International Conference on Structural Health Monitoring and Engineering Structures (SHM&ES 2025)

Table of Contents

1. Optimization and Machine Learning in Engineering Problems

   1. Frontmatter

   2. Optimized Supervised Machine Learning for Accurate Estimation of Reinforcement in RC Beams and Columns

      Nhan Thanh Vu Nguyen, Chon Tran, Duong Thai Le, Quy Thue Nguyen

      Abstract

      In the era of Industry 4.0, technological advancements are transforming the construction industry through automation, artificial intelligence (AI), and data-driven decision-making. Traditional structural design methods, particularly for reinforced concrete beams and columns based on the Vietnamese Standard TCVN 5574:2018, involve multiple manual calculations that, while effective, are time-consuming and labor-intensive. To address this limitation, this study proposes a Supervised Machine Learning (SML) approach to optimize reinforcement design for beams and columns. Using available datasets, the SML models can predict the required reinforcement area with high accuracy, achieving deviations of less than 10% for beams and 13% for columns. The application of SML in reinforcement estimation significantly reduces the time required for structural calculations. Moreover, it lays the foundation for future developments in automated structural design processes through seamless integration with architectural and structural design software and programming environments such as REVIT, ETABS/SAP2000, and MATLAB.

   3. Optimizing a 26-Story Truss Tower Using the K-Means Optimizer Algorithm

      Hoang-Le Minh, Tran Minh Luan, Thanh Cuong-Le

      Abstract

      This study introduces a new optimization method called K-means Optimizer (KO). The special feature of this algorithm lies in the combination of K-means clustering to determine the centroid vectors—representing the regions with high potential in the search space for solutions. Based on those centroids, the algorithm applies two flexible moving strategies, allowing it to both explore new regions and effectively exploit known regions, in order to find the best solution. To evaluate the effectiveness, the KO algorithm is applied to the optimization problem of a 26-storey truss tower structure with a total of 942 bars and 244 nodes, using 59 design variables. Then, the results from KO are compared with two other popular optimization algorithms, ETO and PSO. The results show that the KO algorithm achieves the smallest optimal value and converges faster than the other two methods. Specifically, KO ranks first in performance among the three algorithms, demonstrating its ability to solve optimization problems efficiently. This shows that KO not only performs well on complex models, but is also a reliable choice for engineering problems that require high accuracy and computational efficiency.

   4. Truss Structure Optimization Using the Portia Spider Algorithm: A Bio-inspired Approach

      Vu Hong Son Pham, Thuy Dung Dau, Van Nam Nguyen, Nghiep Trinh Nguyen Dang

      Abstract

      Optimizing truss structures is essential in civil engineering, aiming to reduce weight and support sustainable, high-efficiency designs. In this study, the Portia spider algorithm (PSA) is introduced as a novel optimization algorithm specifically designed for truss design problems with sizing constraints and continuous variables. PSA integrates advanced solution modification strategies to effectively address the inherent complexity of truss structure optimization. The algorithm’s performance was thoroughly evaluated through extensive testing on 25-bar truss structures. The findings indicate that PSA consistently yields superior truss designs compared to other swarm-based techniques, achieving significant weight reductions and enhanced design quality. By offering a robust and computationally efficient solution for truss optimization, PSA demonstrates considerable potential to advance the field of structural optimization. These results highlight PSA as a valuable resource for civil engineers seeking to improve structural performance and efficiency, ultimately contributing to more sustainable construction systems.

   5. A Multiverse Optimizer for Time–Cost Trade-Off of Vehicle Routing Problem

      Vu Hong Son Pham, Van Nam Nguyen, Nghiep Trinh Nguyen Dang, Thuy Dung Dau

      Abstract

      This paper proposes a novel strategy for solving the vehicle routing problem with capacity constraints by applying the Multiverse Optimizer (MVO) algorithm. Inspired by the principles of the multiverse theory, MVO simulates the movement of candidate solutions through metaphorical white holes, black holes, and wormholes to enhance the exploration and exploitation processes. The white hole mechanism supports global exploration, while the black hole and wormhole components help refine and converge toward optimal routes. The proposed method enables a practical trade-off between delivery time and operational costs, making it suitable for real-time logistics planning. A case study involving 20 customer locations illustrates the effectiveness of the approach, achieving a total delivery duration of 4.4 h and an overall cost of $261.59.

   6. An Advanced Metaheuristic Framework for Time–Cost–Quality Optimization in Complex Construction Projects

      Nghia Hoai Nguyen, Khanh-Nhan Tran

      Abstract

      Contemporary construction projects demand a holistic management approach that does not balance time and cost only but preserve quality under uncertain conditions also. Traditional methods often emphasize time–cost trade-offs, overlooking the intricate link between task sequencing and quality outcomes. To address these limitations, this paper introduces a novel optimization framework combining the multi-objective sea-horse optimizer (MOSHO), the root assessment method (RAM), and fuzzy logic, specifically designed for time–cost-quality trade-off (TCQT) scenarios. The proposed model was tested against established algorithms—MOSGO, MOSOS, and NSGA-III on real-world construction data. Results demonstrate that the integrated MOSHO-RAM-Fuzzy logic approach outperforms the conventional mentioned techniques. This consistently generates well-distributed solutions that effectively reconcile cost efficiency, project duration, and product quality. This integrated model also serves as a robust decision-support tool in the context of involving multiple execution alternatives, particularly suited to the scheduling phase of large-scale projects where uncertain parameters significantly impact construction project outcomes.

   7. Evaluation of K-means Optimization Algorithm Using CEC2020 Functions

      Hoang-Le Minh, Tran Minh Luan, Thanh Cuong-Le

      Abstract

      In this study, a new metaheuristic optimization algorithm called K-means Optimizer (KO) is introduced. The highlight of KO is the use of K-means clustering technique to identify center vectors—representing areas with good solution potential in the search space. From these vectors, KO deploys two adaptive migration mechanisms, which help expand the exploration scope while enhancing the exploitation ability in potential solution areas. To verify the effectiveness of the method, the KO algorithm is applied on the CEC2020 benchmark functions set. The results obtained from KO are then compared with three popular optimization algorithms today: SCHO, AOA and GWO. Statistics show that KO achieves more outstanding results in most cases. With outstanding performance and good adaptability to complex problems, KO demonstrates its potential for wide application in engineering and optimal design problems requiring high accuracy and computational efficiency.

   8. Hybrid Machine Learning for Accurate Prediction of CFST Column Compressive Strength

      Tran-Trung Nguyen, Andy Nguyen, Phu-Cuong Nguyen

      Abstract

      This paper introduces a hybrid machine-learning framework to improve the predictive accuracy of ultimate compressive strength in circular concrete-filled steel tube (CFST) columns. The suggested methodology combines CatBoost with Bayesian optimization to enhance model efficacy and computational efficiency. A dataset of 663 experimental specimens is employed for training and validation. Sophisticated data preprocessing methods, encompassing mathematical transformations, are utilized to enhance feature representation. The efficacy of the proposed method is assessed through a comparative analysis with conventional artificial neural networks (ANN). The hybrid CatBoost model demonstrates enhanced predictive accuracy, significantly lowering error metrics compared to ANN-based models. The proposed framework specifically decreases the Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and R² score by 146.55, 262.55, and 0.99%, respectively, illustrating its efficacy in structural engineering applications. The selection of CatBoost is driven by its capacity to manage intricate nonlinear relationships, reduce overfitting, and ensure computational efficiency, rendering it a persuasive alternative to traditional machine learning methods.

   9. Settlement Prediction of Nodular Piles: A Machine Learning Perspective

      Hung La, Tan Nguyen

      Abstract

      Predicting the settlement of nodular piles under static loading is challenging due to the nonlinear nature of pile–soil interaction. In this study, we use a hybrid model that combines Artificial Neural Networks (ANNs) with the Covariance Matrix Adaptation Evolution Strategy (CMA-ES) to tune hyperparameters and network architecture automatically. The model is trained using experimental data that include pile geometry, applied load, and soil conditions. To interpret the model, we apply feature importance method. The results show that the ANN–CMA-ES model produces accurate predictions and identifies the most important input variables, such as load and cylindrical diameter. This modeling approach may help improve decision-making in pile foundation design.

   10. Integrating an Ensemble Machine Learning Model with a Metaheuristic Optimizer to Predict the Compressive Strength of High-performance Concrete Mixtures

       Thuy-Linh Le, Dinh-Nhat Truong

       Abstract

       The complex, nonlinear relationship among the components of high-performance concrete (HPC) poses a significant challenge in modeling its compressive strength. Prior studies have consistently found it challenging to maintain a balance among various factors, including mix proportions, material properties, curing conditions, ambient conditions, and concrete age. Nevertheless, in concrete mix design and quality control, compressive strength remains the primary indicator of HPC quality. This study proposed an ensemble model constructed using a metaheuristic optimization algorithm to predict the compressive strength of HPC. Three datasets are utilized to assess the performance of both single and ensemble models. The optimal results will be further compared with those from prior studies. Analytical results indicate that the proposed ensemble model outperforms others in predicting the compressive strength of HPC.

   11. Investigation of the Effectiveness of Optimization Algorithms in Structures

       Thanh Sang-To, Tan Sy Tran, Thanh Cuong-Le

       Abstract

       In this study, an investigation into the effectiveness of optimization algorithms in structures is presented. A range of modern algorithms, including. Particle Swarm Optimization (PSO), Enhanced Particle Swarm Optimization (EnPSO), Grey Wolf Optimizer (GWO), Salp Swarm Algorithm (SSA), were employed to assess the performance of each candidate in solving optimization problems. The results obtained demonstrate that these algorithms exhibit distinct advanced and disadvanced when addressing specific optimization challenges.

   12. Evaluation of the Exponential-Trigonometric Optimization Algorithm Applied to Truss Structure

       Tran Minh Luan, Minh Thi Tran, Xuan Thinh Nguyen, Thanh Cuong-Le

       Abstract

       This paper introduces a new optimization algorithm called Exponential-Trigonometric Optimization (ETO), inspired by the combination of two important mathematical elements: exponential functions and trigonometric functions. This algorithm is built to achieve a balance between two important stages in the optimization process—that is, exploring the search space and exploiting potential solution regions. By integrating random components and flexible adaptation, ETO shows the ability to find solutions efficiently and avoid local extremes. To verify its effectiveness, the algorithm was applied to the optimization problem of a 200-bar truss structure. The results from the comparative experiment with other famous algorithms such as PSO, SCA and HHO show that ETO consistently gives the best solution and has higher stability. This proves the great potential of ETO not only in the field of structural optimization but also in many other engineering and scientific fields in the future.

   13. Efficient Resource Leveling in Multi-project Scheduling Environment with an Integrated Mountain Gazelle Optimizer and Opposition-Based Learning

       Vu Hong Son Pham, Thuy Dung Dau, Nghiep Trinh Nguyen Dang, Duc Anh Tuan Le, Le Anh Tran

       Abstract

       Construction enterprises often undertake multiple projects simultaneously, necessitating the efficient allocation of shared resources while ensuring adherence to project deadlines. Addressing this challenge requires advanced optimization techniques to achieve resource balance. This study introduces an improved mountain gazelle optimizer (iMGO) incorporating opposition-based learning (OBL) mechanism to enhance search efficiency and solution diversity. By simultaneously evaluating candidate solutions and their opposite counterparts, iMGO mitigates premature convergence and optimizes the exploration–exploitation trade-off. A construction case study is used to validate the effectiveness of the proposed algorithm, demonstrating its superior performance in achieving optimal resource leveling compared to benchmark algorithms. Experimental results indicate that iMGO not only attains optimal solutions but also exhibits greater stability and consistency across multiple trials. These findings highlight the potential of the developed approach to enhance resource management efficiency in complex multi-project environments.

   14. Predicting Labor Cost Performance Index in Construction Projects Using Explainable AI

       Hung Tran Phi, Nghia Hoai Nguyen

       Abstract

       Labor cost performance index (LCPI) is an important factor of financial control and operational efficiency in construction enterprises. The study presents a model that examines labor cost performance in construction projects based on the explainable artificial intelligence (AI). A historical enterprise data of 212 real construction projects was used, including variables such as revenue, cost, project size, and financial parameters to develop and validate the model. CatBoost was identified as the most suitable prediction algorithm, achieving superior prediction accuracy (R² = 0.9568, RMSE = 0.0073). SHAP analysis shows that financial variables, including project value, total cost, and expected profit, are the most influential factors on LCPI. These results contribute to an LCPI prediction model for construction businesses, aiming to help managers plan finances and select projects strategically with machine learning tools.

   15. Neural-Network Guided Minima Forecasting for an Enhanced Particle Swarm Optimizer

       Tri Ton That, Binh Le-Van, Thanh Cuong-Le

       Abstract

       Particle Swarm Optimization (PSO) is appreciated for its simplicity and ease of adaptation, yet its progress may stall when the swarm fails to identify promising regions early in the search. We present Neural-Network-Predicting Adaptive PSO (NNP-APSO), a variant that embeds an artificial neural network trained online to approximate the lower envelope of the objective landscape. At each iteration the network forecasts a candidate minimum, which is injected into the velocity update of under-performing particles; the swarm therefore shifts its exploration–exploitation balance automatically, without introducing additional control parameters. NNP-APSO is evaluated on the standard single-objective test suite (15 classical benchmarks) and compared, under identical computational budgets, with six established optimizers: Simulated Annealing, canonical PSO, Whale Optimization Algorithm, Grey Wolf Optimizer, Genetic Algorithm and Artificial Bee Colony. The experimental study indicates that NNP-APSO offers a promising alternative, delivering good and noteworthy results while highlighting the potential of real-time neural guidance within the PSO framework.

   16. Seasonal Cash Inflow Optimization in Construction Projects Using the Termite Life Cycle Optimizer

       Hung Tran Phi, Nghia Hoai Nguyen

       Abstract

       Construction cash inflows are characterized by seasonality, which often leads to inefficiencies in resource utilization and financial instability. The goal is to minimize monthly cash flow variability while satisfying resource constraints and financial efficiencies. This paper proposes a novel cash inflow curve flattening framework by integrating Termite Life Cycle Optimizer (TLCO), a bio-inspired metaheuristic algorithm, to optimize project selection by flattening the cash flow curve. The results show that TLCO performs well in this problem, with good experimental results and better convergence than traditional algorithms. The study combines optimization with 03 simulation scenarios on investment in analytical technology, resulting in visualization with high positive impact. The proposed model provides construction companies with a powerful analytical tool to support decisions to stabilize cash flows, reduce dependence on external financing, and support long-term sustainability.

   17. Efficient Design of Single Mooring Buoy Lines: A MOMSA-Based Approach

       Quang Thanh Do, Quoc Hoan Pham, T. Vu-Huu, Thanh Cuong-Le

       Abstract

       Designing mooring lines for floating buoys is essential to keep them stable and secure, especially in open water with unpredictable waves, wind, and currents. This paper introduces a practical way to improve mooring line design by using catenary theory to model how the line behaves and applying the Multi-objective Mantis Search Algorithm (MOMSA) for optimization. The aim is to find mooring line solutions that use less material and reduce how far the buoy drifts from its intended position. The approach also considers real-world requirements, such as choosing line weights from standard product lists and following national technical standards. The results show that MOMSA can successfully provide engineers with various good design options, helping them choose between safety and material costs.

   18. Prediction of Concrete Compressive Strength Using Boosting-Based Machine Learning Algorithms

       Truong-Giang Nguyen, Van Than Tran, Thanh Danh Tran

       Abstract

       This study investigates the use of boosting-based machine learning algorithms to predict the compressive strength of concrete, aiming to improve structural safety, optimize mix proportions, and enhance construction efficiency. Traditional empirical models often fall short in modeling the complex nonlinear relationships among input materials. Five popular boosting methods—AdaBoost, Gradient Boosting Machine (GBM), XGBoost, LightGBM, and CatBoost—were evaluated using a benchmark dataset of 1030 samples from the UCI repository, containing eight numerical features. Model performance was measured using the coefficient of determination (R²). Among the methods, CatBoost outperformed others with R² = 0.9943 on the training set and 0.9440 on the testing set, followed by XGBoost and GBM. AdaBoost showed the weakest performance. The results highlight the strong capability of advanced gradient boosting algorithms, particularly CatBoost, in modeling the nonlinear behavior of concrete materials.