Taiwan Association of Engineering and Technology Innovation: E-Journals
Not a member yet
887 research outputs found
Sort by
Improving Cardiac Computed Tomography Scan Segmentation Using a U-Net Model with Continual Learning Techniques
Full text linkAccurate segmentation of cardiac structures in computed tomography (CT) scans is challenging due to the proximity and similar intensity of adjacent organs. This study introduces an enhanced U-Net-based approach incorporating continual learning, class merging, and separation strategies to improve cardiac CT segmentation. Anatomically related structures are first merged and later separated through class-specific heads, reducing boundary misclassification. Furthermore, pixel adjacency is employed to improve the delineation of complex cardiac regions. The proposed method is evaluated on the MM-WHS 2017 dataset, focusing on seven components: left ventricular cavity (LVC), right ventricular cavity (RVC), left atrium cavity (LAC), right atrium cavity (RAC), myocardium (MYO), ascending aorta (AA), and pulmonary artery (PA). Experimental results show that the proposed model achieves a dice score coefficient (DSC) of 94.08% and an intersection over union (IoU) of 92.03%, outperforming baseline U-Net models. These findings demonstrate the effectiveness of structure-aware learning in advancing cardiac CT segmentation
Improving Solar Energy Reliability with Data-Driven Anomaly Detection Techniques
Full text linkThis study investigates unsupervised machine learning (ML) for anomaly detection in solar photovoltaic (PV) power generation data from 2019 to 2023. An unsupervised approach is selected to overcome the absence of pre-labeled fault data, enabling the autonomous identification of operational patterns. Following data preparation, K-means clustering (k=3) identifies distinct operational patterns, specifically characterizing regimes such as optimal performance (Cluster 2) and low energy output attributed to adverse weather conditions (Cluster 1). These clusters are subsequently visualized using principal component analysis (PCA) to validate their distinct separation. An isolation forest model is then employed for anomaly detection, identifying 17 significant deviations. These anomalies occur most frequently in 2020, coinciding with the COVID-19 pandemic period. Many fall outside the typical energy range of 2.0–3.2 kWh/day and are associated with non-ideal weather conditions. This finding demonstrates that unsupervised ML provides a scalable framework for monitoring PV system health, enhancing reliability, and supporting preventive strategies
Multi-Target Particle Swarm Optimization with Machine Learning Surrogates for Efficient Concrete Mix Design
Full text linkThis study presents a multi-target particle swarm optimization (MT-PSO) approach for efficient concrete mix design. It simultaneously designs mixes with multiple predefined strengths under a constant water-cement ratio. A gradient boosting-based surrogate model, trained on experimental mix data, predicts compressive strength. The modified particle swarm optimization (PSO) algorithm accommodates multiple targets in parallel, allowing solution sharing across target groups. MT-PSO is compared with a repeated PSO (R-PSO) strategy that optimizes each target separately, both minimizing the absolute error between predicted and desired strengths. Across 30 independent trials, MT-PSO consistently achieves lower mean errors, smaller deviations, and faster convergence, often reaching R-PSO’s final accuracy within only a few iterations. Moreover, MT-PSO requires over 85% fewer fitness evaluations. These results demonstrate the superior accuracy, robustness, and computational efficiency of MT-PSO for multi-target optimization problems
Ranking of Security Factors in Blockchain-Based IoT Paradigm Using AHP-TOPSIS Method
Full text linkThe Internet of Things (IoT) connects smart devices for efficient data sharing but faces challenges such as low processing power, limited storage, and security risks. Blockchain technology offers a secure and decentralized solution to these issues. This study prioritizes key security factors in blockchain-based IoT systems using a hybrid approach combining the Analytic Hierarchy Process (AHP) and the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS). AHP determines the weights and relative importance of security factors, while TOPSIS ranks them based on closeness to the ideal solution. The results demonstrate a strong Spearman correlation (ρ = 0.9916) with individual AHP and TOPSIS outcomes, and sensitivity analysis confirms the stability of the rankings. Data integrity, access control, and authentication are identified as the top three security factors. These findings support the development of secure and scalable blockchain-based IoT systems
Effect of Weir Height on Undershot Water Wheel Turbine Performance
Full text linkUndershot water wheel turbines are suitable for operation in low-elevation regions and locations with very low water head. Turbines with curved blade configurations are commonly implemented to optimize efficiency. This research examines how variations in weir height influence the performance of undershot turbines when operating under limited water flow conditions. A turbine with six blades is tested experimentally and numerically using seven weir heights ranging from 0.02 m to 0.14 m. The study used both experimental and computational fluid dynamics approaches, employing the dynamic mesh model. The results indicate that applying a curved weir as a passive flow-control device can significantly enhance turbine efficiency. At a weir height of 0.10 m, the maximum efficiency reached 80.83% based on the experimental approach and 84.69% based on the computational approach. These findings demonstrate the potential of weir-assisted undershot turbines for energy harvesting in shallow rivers under low-flow conditions
Examining the Impact of Sustainable Value and Economic Value on TPASS Usage Intention
Full text linkAs Taiwan faces the challenge of reducing per capita carbon emissions from 12 tons per year—twice the global average—to achieve the “net zero by 2050” goal, understanding the factors influencing the adoption of sustainable transportation becomes crucial. This study examines how sustainable value and economic value affect TPASS usage intention, using an integrated model based on the value-attitude-behavior theory and the theory of reasoned action. The model investigates how perceived economic value, perceived sustainable value, along with attitude and subjective norms, shape consumers’ intentions to use TPASS commuter passes. Data are collected through online surveys, yielding 302 valid responses, and analyzed using structural equation modeling. Results show that both sustainable and economic values significantly influence TPASS attitude, with economic value having a stronger effect. Furthermore, attitude and subjective norms both positively affect usage intention, with subjective norms demonstrating a notably stronger impact, providing insights for sustainable transportation policies
A Statistical Approach to Evaluating the Influence of Paving Block Thickness and Sand Bedding on Surface Deflection
Full text linkThis study aims to investigate the structural behavior of concrete paving block (CPB) road surfaces. A statistical approach was used to analyze surface deflection using a light weight deflectometer (LWD). To assess the statistical significance of each factor’s influence, analysis of variance (ANOVA) was conducted. To analyze the relationships between variables, correlation and linear regression analyses were performed. The results show that increasing the thickness of the CPB and sand bedding significantly reduces deflection, while greater deflection occurs with higher loads. As the thickness of the CPB and sand bedding increased, the measured deflection decreased from 0.446 mm to 0.259 mm. Correlation analysis shows a negative correlation between deflection and thickness or strength, and a positive correlation with load. ANOVA analysis confirms that all of the variables significantly affect deflection. The regression equation that includes all of the variables and uses load as a multiplication factor is the most accurate
Dynamic Response and Shear Mechanisms of Reinforced Concrete Columns Subjected to Lateral Impact
Full text linkThis study investigates the dynamic response and shear failure mechanisms of reinforced concrete (RC) columns under lateral impacts, such as vehicle collisions. A three-dimensional nonlinear finite element (FE) model is developed, incorporating concrete damage, reinforcement plasticity, and strain-rate effects. The numerical model is validated against experimental pendulum impact tests. Parametric analyses are conducted to evaluate the effects of impact severity, axial compression ratio, stirrup ratio, longitudinal reinforcement ratio, slenderness ratio, and concrete strength on failure modes. Results indicate that impact-induced diagonal cracking governs the transition from non- failure to punching-shear failure. Increasing the stirrup ratio and concrete strength delays brittle shear failure, whereas excessive axial compression promotes shear localization. The findings provide insights into the impact-resistant behavior of RC columns and offer guidance for improving structural design against lateral impact loads
Analysis and Selection of the Best Global Geopotential Model for Lebanon: Case Study Rashya District
No full textAccurate transformation of Global Positioning System (GPS) derived ellipsoidal heights to orthometric heights necessitates the selection of an optimal Global Geopotential Model (GGM). This study aims to identify the most accurate freely available high-resolution GGM for Lebanon. The performance of five GMs, Earth Gravitational Model 2008 (EGM2008), SGG_UGM_1, SGG_UGM_2, GECO, and XGM2019e, is assessed. Geoidal undulation values are extracted for 28 geodetic benchmarks in the Rashaya district using a Geographic Information System and compared with reference data from the Lebanese Directorate of Geodetic Affairs. Vertical accuracy is quantified using the mean deviation (DN), standard deviation, and root mean square error (RMSE) between GGM-derived and reference heights. The analysis reveals that XGM2019e provides the highest accuracy, with the smallest mean deviation (DN = –0.25 m) and the lowest RMSE (±1.09 m). These results establish XGM2019e as the optimal GGM for Lebanon, ensuring precise height transformation and supporting advanced geodetic and geospatial analyses
A Parallel Prediction Method for Battery Capacity Based on the Multiscale and Temporal-Spatial Feature Fusion
Full text linkThis study proposes a parallel feature fusion-based method to accurately predict battery capacity for battery health management (BHM). Existing data-driven approaches suffer from ineffective extraction of multi-scale features and feature redundancy induced by sequential strategies. To address these gaps, a hybrid deep learning framework is proposed. Specifically, discharge voltage data are first standardized to unify sample dimensions. Then, parallel multi-scale branches are constructed to simultaneously capture the spatial and temporal features of battery discharge signals. A channel attention module is subsequently employed to adaptively filter redundant features and enhance the weight of degradation-related feature representations. Finally, a fully connected network maps the refined features to battery capacity values. The experimental results validate that the proposed method outperforms single-model and sequential baselines, with 0.00369 to 0.01331 RMSE on NASA battery, 0.0498 on CALCE batteries, and 1.8196 on Oxford batteries