Engineering Journal (Faculty of Engineering, Chulalongkorn University, Bangkok)
Not a member yet
1223 research outputs found
Sort by
Enhancing Fault Diagnosis in Imbalanced Data Using Weighted GRU Architecture
No full textThe class imbalance, characterized by an unequal distribution between normal and abnormal classes, is predominantly observed in the field of fault diagnosis. Abnormal classes typically represent a minority, leading to a biased learning process favoring the majority class. Therefore, class balancing techniques are essential when applying deep learning approaches to ensure accurate classification of minority fault classes. In this study, we investigate and propose weighted approach for the gated recurrent unit (GRU) algorithm. The proposed weighted approach adjusts all three weights-input, recurrent, and bias inside the GRU architecture. Additionally, the synthetic minority over-sampling (SMOTE) technique with vanilla GRU and long short-term memory (LSTM) as well as the combination of SMOTE and the proposed weighting technique for GRU and LSTM, are compared to the proposed weighting architecture with GRU. We evaluate the effectiveness of this technique using operational data from a real multi-stage flash desalination plant, synthesizing datasets with varying imbalance ratios (4, 9, and 14) for evaluation. Performance metrics such as accuracy is employed for evaluation. Among the models tested, the weighted GRU (WGRU), the proposed model, consistently outperforms others across all variables and imbalance ratios
Implementing a Six-Element Framework of Safety Culture in the Thai Cosmetics Industry
No full textThis study investigates the implementation of safety culture systems in the cosmetics industry using a structured six-element framework: Input, Processing, Output, Working Environment, Ergonomics, and Safety Experience. A mixed-methods approach was employed to capture both quantitative and qualitative insights, highlighting significant improvements in workplace safety, operational efficiency, and regulatory compliance with GMP and ISO 22716 standards. Quantitative findings revealed statistically significant advancements across all six elements, with notable gains in resource allocation and the integration of safety protocols. Qualitative data identified leadership commitment and tailored training programs as critical facilitators of success, whereas technological constraints and cultural resistance were recognized as key implementation challenges.
In response to these barriers, the study emphasizes the importance of scalable and adaptive technological solutions. Specifically, it proposes the future integration of Industry 4.0 technologies—such as IoT-enabled monitoring systems and AI-driven analytics—to enhance hazard prediction, streamline safety management, and support real-time decision-making. These findings provide a forward-looking roadmap for advancing safety culture practices not only in the cosmetics sector but also across similarly structured manufacturing industries
Seismic Strengthening of Partial Infilled RC Frame with Upper Opening Using Ferrocement and Expanded Metal Mesh
No full textSevere damage often occurs in typical low-rise commercial buildings with reinforced concrete (RC) infilled frames featuring openings on the ground floor during earthquakes. Common failure modes include shear failure due to short-column behavior, as well as flexural and shear failures at the ends of columns and beams. To improve the shear and flexural strength of beam-column joints and enhance the lateral strength and ductility of infilled masonry walls, this study proposes a strengthening method for partial infilled RC frame with upper opening. An analytical model was also developed to predict the lateral strength and ductility of the frames. This research investigates the strengthening behavior of partial infilled RC frame with upper opening, reinforced using ferrocement and expanded metal mesh. The specimens were subjected to constant vertical loads and cyclic lateral loads. The experimental study involved two full-scale, single-story and single-bay frames: (1) a control specimen with an upper panel opening (UPF-C) and (2) a specimen with an upper panel opening strengthened using ferrocement and expanded metal mesh (UPF-S).The results demonstrate that the UPF-S specimen exhibited greater lateral resistance, stiffness, and ductility compared to the control specimen (UPF-C). The strengthening method effectively mitigated damage to the RC infill frame by shifting the failure behavior of beam-column joints from shear failure to ductile failure. Finally, the experimental results were analyzed and compared with nonlinear analytical models. The proposed model yielded predictions closely aligned with the experimental findings, confirming its reliability and consistency
PVSyst Enabled Real-Time Techno-Economic Assessment of a 1.5 MWp Grid-Tied Solar Photovoltaic System
Full text linkPakistan is facing an escalating energy crisis that demands immediate and sustainable solutions. To address this challenge, this study evaluates the feasibility of solar energy production through a techno-economic analysis of a 1.5 MWp photovoltaic system. The analysis combines PVsyst simulations with real-time operational data collected between June 2023 and January 2024, enabling a direct comparison between predicted and actual system performance. The study examines energy yield, performance ratio, and financial metrics to assess economic viability. Results reveal that the average performance ratio was 82.7% in simulation but improved to 88.05% in real operation, reflecting a 5.35% gain in energy output. Financial analysis shows that the system achieves payback within 6.8 years, with a levelized cost of energy of 3.43 PKR/kWh and levelized savings of 13.23 PKR/kWh, amounting to cumulative lifetime savings of 898.2 millions PKR. The findings highlight the reliability and profitability of solar PV in Pakistan and demonstrate that integrating simulation and real-time data provides a more accurate basis for investment decisions. This study contributes a practical framework for bridging the gap between theoretical modeling and operational outcomes, offering actionable insights for accelerating the transition to economically viable green energy solutions
ECGNet-ViT: Hybridizing GoogleNet with Vision Transformer for Accurate COVID-19 Detection from ECG Images
Full text linkCOVID-19 has affected millions of people around the world in the last three years. Despite widespread vaccination efforts, infections persist and definitive treatments remain elusive. Therefore, early and accurate detection of COVID-19 is critical to minimize invasive procedures and reduce mortality. Although radiographs and CT scans are commonly used for the diagnosis of COVID-19, electrocardiogram (ECG) images remain underutilized despite their widespread availability. This limited use can be attributed to the complex transformations required to process ECG data, which increase computational demands. This study proposes a novel hybrid deep learning model ECGNet-ViT for COVID-19 detection. The model combines the multi-scale feature extraction capabilities of GoogleNet (GNet) with Swish activation functions and densely connected layers, and then integrates it with Vision Transformer (ViT) to effectively capture long-range dependencies in classification tasks. This approach can efficiently analyze ECG data and accurately classify samples into five categories: normal, COVID-19, myocardial infarction (MI), previous myocardial infarction (PMI) and arrhythmia (AHB). Comprehensive experiments on a publicly available ECG datasets demonstrate the effectiveness of the proposed model, achieving 99.13% accuracy, 99.19% precision, 99.24% recall, and 99.22% F1 score. These results highlight the potential of the proposed model to provide reliable, non-invasive support in COVID-19 diagnosis based on ECG data
Performance Improvement for Germanium-Based Near-Field Thermophotovoltaic Converter
No full textNear-field Thermophotovoltaic (NF-TPV) converter utilizes the tunnelling of an evanescent wave to surpass the blackbody limit, enhancing the radiative heat transfer of the TPV converter by several orders of magnitude. One of key challenges for commercial NF-TPV converter is cost reduction. Germanium is known to be a cost-effective material with bandgap energy compatible with a TPV application. In this study, the germanium-based NF-TPV converter is introduced with an addition of air-bridge gap (ABG) at the interface between a TPV cell substrate and a metal back surface reflector (BSR) as a strategy to improve sub-bandgap photon utilization. Effects of air-bridge gap and germanium substrate thickness on radiative heat fluxes and converter performance are investigated. At the radiator temperature of 1400 K and the optimum air-bridge gap thickness of 1300 nm, system efficiency of NF-TPV converter increases from 17.16% to 24.98% for a 175 thick TPV cell, and from 16.73% to 31.39% for a 40 thick TPV cell under moderate surface passivation. In addition, the converter's performance under varying radiator temperature is analysed. This study demonstrates the potential of the air-bridge gap to optimize NF-TPV converter performance
Characterization of Metal Fumes and Oxides Across Current Variations in the Shielded Metal Arc Welding (SMAW) Process
No full textThis study explores the formation of metal fumes during the Shielded Metal Arc Welding (SMAW) process using SS400 low carbon steel, 6 mm in thickness, in accordance with JIS G3101 and AWS D1.1/D1.1M standards. The research was conducted across a range of electrical currents, from 60A to 130A. To analyze weld profiles, fume morphology, elemental composition, and metal fume concentrations, scanning electron microscopy coupled with energy-dispersive X-ray spectroscopy (SEM-EDS), X-ray diffraction (XRD), and inductively coupled plasma mass spectrometry (ICP-MS) were utilized. The results revealed that the lowest concentration of metal fumes, measured at 3.961 mg/m³, occurred at a welding current of 70A, with iron (Fe), zinc (Zn), and nickel (Ni) as the predominant elements. Conversely, the highest fume concentration of 18.483 mg/m³ was found at 110A, characterized by elevated levels of iron (Fe), manganese (Mn), and nickel (Ni). The analysis also identified the presence of various metal oxides, including iron oxides (hematite and magnetite), titanium dioxide (rutile and anatase), silicon dioxide (SiO2), and copper (II) oxide (CuO). These findings highlight the importance of selecting appropriate welding currents, particularly 70A and 90A, to achieve optimal weld quality while reducing exposure to hazardous fumes
Smart Microscopy Camera Kit: Automatic Counting of Blood Cells in Peripheral Blood Smear Images Using RetinaNet on Raspberry Pi CM3+
No full textMicroscopic examination of peripheral blood smear images for blood cell counting remains a critical yet labor-intensive task in clinical diagnostics. This research presents MicrosisDCN, an intelligent microscopy camera system designed to automate blood cell detection and counting, powered by a compact embedded platform based on the Raspberry Pi Compute Module 3+. The system incorporates a 5-megapixel image sensor and a versatile eyepiece fitting that is compatible with the most compound microscopes, providing a portable, cost-effective, and user-friendly solution. Calibration procedures ensure alignment with traditional high-power field (HPF) standards, allowing cell counts to be reported in standard mitotic count units. To detect red blood cells, white blood cells, and platelets in real-time, the system uses a special version of a deep learning model called RetinaNet, which has been improved with a technique called auto-anchor parameterization. MicrosisDCN achieves a mean Average Precision (mAP) of 86.81% in detecting a few types of blood cells with minimal errors: 1.06% for red blood cells, 0.06% for white blood cells, and 4.23% for platelets. The results indicate that MicrosisDCN, which combines traditional microscopy with advanced vision technologies, serves as an efficient, practical, and scalable solution for clinical and medical laboratory applications
A Fuzzy Cognitive Map for Identifying the Interrelationship of Critical Success Factors for Digital Transformation Implementation by Construction Contractors
No full textDigital transformation (DX) is imperative for every business entity that wants to be competitive in its industrial sector. The rapid rise and widespread propagation of the coronavirus (COVID-19) has hastened DX across various industries, especially the construction industry. As a result, construction companies are on the verge of radical DX, which is primarily driven by digital innovation and advancement. However, knowledge regarding DX implementation for construction firms is limited and fragmented, leading to unsuccessful DX implementation for most of the construction companies. This paper investigates the interrelationship of the critical success factors (CSFs) for DX implementation by construction contractors. A fuzzy cognitive map (FCM) is adopted to analyze the interrelationship of these factors, constructed based on inputs from 17 experts with extensive experience in construction and digital technologies. The outdegree, indegree, and total degree values of the FCM indicate the factors that influence other factors the most, the factors that are influenced by other factors the most, and the most important factors, respectively. It is found that effective leadership has the highest outdegree value and total degree value, whereas the research and development capability of organizations has the highest indegree value. These findings provide the fundamentals of DX implementation in construction businesses and their interrelationship. For example, understanding that effective leadership strongly influences other success factors enables firms to prioritize leadership development initiatives early in the DX process. Likewise, recognizing that research and development capability is highly influenced by other factors helps contractors identify which upstream capabilities must be strengthened first. By leveraging these insights, construction contractors can embrace digital innovations and technologies for their businesses successfully
Optimization of Irrigation in Open-Field Fruit Orchards Using an Intelligent Precision Control System: A Case Study of Durian
No full textPrecision irrigation systems (PIS) are essential for optimizing water use, crop yield, and fruit quality. Estimating crop water requirements has become increasingly complex due to growing variability in weather patterns, which increases the risk of irrigation mismanagement. Conventional PIS often rely on fixed schedules or pre-determined water quantities. While some offer decision-support capabilities, they typically require human intervention and lack the ability to control irrigation. To address these issues, an intelligent precision irrigation system (IPIS) was developed. The IPIS utilizes real-time meteorological data collected from local sensors and implements advanced control algorithms based on the Penman-Monteith evapotranspiration method to optimize water delivery. A four-month field experiment with 166 irrigation events conducted in a durian orchard in Rayong, Thailand, demonstrated the system's ability to dynamically adjust irrigation schedules and water volumes in response to fluctuating weather conditions while maintaining optimal soil moisture levels. This resulted in significantly improved irrigation accuracy and crop water use efficiency. The findings suggest that incorporating machine learning and artificial intelligence in future iterations could further enhance the system’s adaptability, autonomous operation, and predictive capacity, advancing its application in precision agriculture