Engineering Access

Transformer Maintenance Strategies: A K-Means Based Approach for 33 kV DTs

Kittisak Chaisuwan — 2025-06-30

The distribution transformer (DT) is crucial for connecting utility providers to consumers, and its failure can disrupt the distribution network's reliability. The Provincial Electricity Authority (PEA) in Thailand manages a large number of transformers, necessitating efficient maintenance planning to prevent DT failures. This paper introduces a method for classifying the condition of 33 kV DTs without pre-existing cluster data, utilizing the K-means clustering algorithm on data from 150 samples. The dataset includes 7 features from DT annual maintenance records and the Geographic Information System (GIS) of PEA Southern Area 3. Key factors identified are insulation between high voltage and ground, high-low voltage, and low voltage-ground. The method categorizes DT conditions into three clusters: "poor," requiring urgent action; "risk," requiring close monitoring; and "normal," requiring routine maintenance. Validation with K-Nearest Neighbors yields an accuracy of 96.67%, demonstrating the effectiveness of the proposed classification method.

Detection of Parkinson’s Disease Using Voice and Spiral Drawings on Machine Learning Approaches

P. Bhuvaneswari — 2025-06-30

Parkinson's disease (PD) is a neurologically development disease that beginnings with a mild quivers in limbs and affect firmness of the body. Over 6 million individuals around the globe are affected due to this disease. It’s very hard to identify PD in the early stages as there is no particular analysis for this condition and limited expert specialists at present. This recommended procedure for diagnoses of PD by using spiral drawings and speech taken from the patient using eXtreme-Gradient Boosting (XGBoost) algorithms and Random-Forest algorithm pilot to higher efficiency and higher than 74%. The speech samples of the person are studied to identify this sickness. The data from the patients affected by this disease, and non effected people are considered while training this algorithm. From the dataset 40% of voice data is utilized for testing the model, and 60% of spiral data is utilized for training the model. The speech data is represented in 24 columns, represents the status of the person either diseased or healthy. These factors are considered in identifying the disease where 1’s in status column indicate affected person and 0’s indicate healthy person. Mean, standard deviation, jitter, noise to harmonics and harmonics to noise ratio are some of the parameters taken into consideration for predicting the disease for the speech data.

Evaluating the Performance of YOLO Architectures for Effective Gun and Knife Detection

Suradej Intagorn — 2025-06-30

In recent years, the rise in mass shootings in Thailand has highlighted the need for more comprehensive and cost-effective security solutions. One approach is using artificial intelligence to assist human security personnel, particularly for weapon detection through security cameras. Although advancements in deep learning and computer vision have made it possible to deploy such systems on edge computing devices, real-time weapon detection still faces challenges like accuracy and latency. This study addresses the gap in weapon detection research specific to Thailand by utilizing a dataset featuring local environments and weapons, which differ from those in existing datasets. We compare the performance of YOLO versions 5 through 8, focusing on their mean average precision (mAP) in detecting guns and knives. Since each YOLO version is developed by different research teams and may perform differently under specific conditions, our evaluation considers these variations. The findings indicate that YOLOv8 achieves the highest mAP, with scores of 0.874 on the validation set and 0.848 on the test set, demonstrating its effectiveness in the Thai context.

Application of PCLake Model to Predict Water Quality in Tropical Reservoirs, a Case Study of Khlong Luang Ratchalothorn Reservoir

Pongsakorn Wongpipun — 2025-06-30

This study addresses a previously unexamined water quality issue in the Khlong Luang Ratchalothorn (KLR) reservoir, which suffered severe eutrophication in 2018 due to the cyanobacteria Microcystis aeruginosa. This phenomenon led to the development of green scum, alterations in color, foul odors, and diminished dissolved oxygen levels, negatively impacting the reservoir's usability. Utilizing the calibrated and validated PCLake water quality model, the research assessed the impact of varying reservoir inflow scenarios—minimum, average, and maximum—based on rainfall data from 2013 to 2022 on Chlorophyll a (Chla) levels. Additionally, the effects of reducing nutrient and suspended solids loading by 10%, 30%, and 50% were evaluated under worst-case scenarios. Findings indicate that the reservoir maintains an eutrophic status across drought, normal, and wet conditions, with drought having the most significant adverse impact on water quality, potentially persisting for up to two years. Despite a 50% reduction in nutrient and suspended solids, the reservoir does not transition to a mesotrophic state; however, water users do not face significant challenges even under these extreme conditions. This research offers new insights into effective strategies for managing water quality in similar aquatic systems.

Effects of Mixing Techniques on Properties of Reactive Powder Concrete Containing Fly Ash

Jinhao Wei — 2025-06-30

Reactive powder concrete (RPC) is an ultra-high-performance concrete with exceptional mechanical properties and durability compared to conventional concrete. Due to its unique properties, RPC is a promising material for advancing sustainable, resilient, and eco-friendly infrastructure solutions on a global scale. Using RPC, infrastructures can achieve enhanced characteristics such as reduced cross-sections, extended spans, intricate design, and decreased maintenance requirements. Despite its immense potential, RPC remains in the early stage of development, with its properties highly influenced by various factors including ingredient selection, proportions, mixing techniques, and curing method. Typically composed of cement, silica fume, fine sand, superplasticizer, water, and small steel fiber, RPC features a very low water-to-cement (w/c) ratio and densely packed particles. Fly ash, a byproduct of coal combustion, offers a way to enhance RPC properties, while simultaneously reducing costs and environmental impact by substituting for a portion of cement. However, the challenges associated with mixing RPC, especially with low w/c ratios, remain significant, and studies on the effects of mixing techniques for RPC containing fly ash are limited. This study investigates the effects of mixing techniques on the properties of RPC containing fly ash. Specifically, it examines the impact of single-batching and double-batching methods, as well as the influence of different steel fiber types (single-size and mixed-size) and mixing temperatures (25°C and 40°C). Both fresh and hardened properties of RPC were considered. Results indicate that the double-batching method significantly improves the properties of RPC compared to single-batching. While mixed-size fibers enhance the 28-day compressive strength, single-size fibers yield a higher flow value. Higher mixing temperatures negative impacted workability. The findings suggest that optimizing the batching method, fiber type, and mixing temperature can improve both the mechanical properties and workability of RPC.

Optimizing Scheduling of Ready-Mixed Concrete Trucks from Multiple Plants Using Whale Optimization Algorithm

Kittipong Thawongklang — 2025-06-30

Concrete plants face a challenging scheduling problem, as concrete from multiple plants must be efficiently delivered to various construction sites using different fleets of vehicles. This logistical challenge requires optimizing the dispatching schedules of ready-mixed concrete (RMC) plants. Using proprietary urban traffic data from Udornthani, a major city in Northeastern Thailand, this research aimed to identify the most suitable algorithm for optimizing the dispatching schedule under multi-plant and multi-site operations, using concrete waiting time as a key performance indicator. The study compared simulation results with current practices to analyze factors influencing dispatch efficiency, particularly focusing on trucks. Heuristic techniques, which offer quick solutions using simple rules, were employed. The Whale Optimization Algorithm (WOA) was selected for its high efficiency in solving complex problems. Results showed a significant reduction in median waiting times, from 17 minutes to zero, with a p-value < 0.001. WOA improved scheduling efficiency by reducing waiting times by 40-100%, outperforming the manual calculations performed by dispatch officers.

The Impacts of Truck Vibration on the Communities Along the Highway Road No 12, Phitsanulok

Kampanad Hlongjino — 2025-06-30

The purpose of this research is to analyze the vibration effect from different types of trucks passing through interprovincial highway road No. 12 to the community areas. The Peak Particle Velocity and frequency of Zero crossing and Fast Fourier Transform of an individual truck running on a highway road were obtained by the vibration meter. In the first experiment, vibration and speed data were collected from 238 identified trucks weighing from 5 to 50.5 tons and speeds ranging from 51 to 80 km/hr. All these experiment trucks run through the highway balance station regarding weight inspection with camera records. The second experiment was conducted by installing the vibration meter on two buildings located by the roadside in the community area, which is adjacent to the traffic light. The results showed that the vibrations generated by trucks were dominated by vertical waves and Peak Particle Velocity was correlated with the weight and speed of trucks. Interestingly, significant differences in vibration data between two types of trucks, semi-lorry and full-lorry trucks, were found. The impact of vibration on the community was minimized following the Deutsches Institut fur Normung E.V. (DIN) standard which had no significant effect on buildings and the vibration perception was at a level where it was possible to feel the vibration when compared to the guidelines by Whiffin and Leonard.

Effects of Sequence Mixing of Stereocomplex Poly(Lactic Acid)/Propylene-Ethylene Copolymer Blends for Plastic Packaging

Jetawat Wadputi — 2025-06-30

This work examines the effect of sequence mixing stereocomplex poly(lactic acid) (Sc-PLA) on the thermal and mechanical properties of Sc-PLA/elastomer blends. In this study, PLA was melted and blended with poly(D-lactic acid) (PDLA) in an internal mixer at the ratios of 95/5, 90/10, and 85/15 by weight, and then the sample was molded by compression molding. The most significant improvement was found in blending 10 wt% PDLA. The results show the highest Young’s modulus at 2200 MPa from the neat PLA at 1800 MPa. The stereocomplex PLA/PDLA (Sc-PLA/PDLA) at proportion 90/10 wt% by being chosen blended with propylene-ethylene copolymer (PEC) elastomer to study the effect of sequence mixing of Sc-PLA/PDLA/elastomer blends on thermal and mechanical properties and the amount of PEC elastomer was fixed at 10 wt% in the mixture. The mechanical properties show results at 10% break elongation of Sc-PLA/PDLA/PEC blends that is higher than the others, which confirmed that PEC elastomer, which is improved features of Sc-PLA/PDLA flexibility blends and sequence mixing of PLA stereocomplex/elastomer blends affected to the mechanical properties. The result of thermal properties shows the sequence mixing of PLA stereocomplex/elastomer blends does not affect the melting temperature but affects the enthalpy (∆H_m-∆H_c) by Sc-PLA/PEC add PDLA significant decreasing at 23 J/g °C by the other conditions show about 28 J/g °C. It supports the sequence of mixing at Sc-PLA/PDLA/PEC mixed simultaneously, which is the best. The analysis results show that sequence mixing significantly affects polymer blends' properties, which can be applied to industrial processes.

Experimental Investigation and Development of Fuzzy Logic-Based MPPT for Photovoltaic Systems Across Varied Climatic Conditions

P. Poornima — 2025-06-30

The Fuzzy-based photovoltaic (PV) Maximum Power Point Tracking (MPPT) algorithm is a sophisticated approach for enhancing the efficiency and performance of solar photovoltaic systems. It uses fuzzy logic principles to dynamically track and maintain the optimal operating point related to PV panel, ensuring that the determined available power is extracted under varying environmental conditions. Unlike traditional MPPT techniques, the Fuzzy-based PV MPPT algorithm excels which adjusting to changing rapidly the weather surroundings and is more robust in partial shading scenarios. It employs linguistic variables and rule-based decision-making to continuously adjust the voltage which related to the current at which the PV panel operates. This enables the system to efficiently respond to factors such as cloud cover, shading, and temperature variations, optimizing energy production and reducing energy losses. The adaptability and robustness of the Fuzzy-based PV MPPT algorithm make it a valuable tool for harnessing renewable solar energy, contributing to sustainable power generation and reducing reliance on conventional energy sources. The Fuzzy-based MPPT algorithm presents a compelling solution to address the shortcomings of conventional MPPT controllers and increase 2. operational effectiveness of solar power systems.

Assessment of Suspended Sediment Concentration in the Mekong River Using Landsat-8 Data

Jurawan Nontapon — 2025-06-30

The Mekong River is a vital waterway in Southeast Asia, significantly impacted by sediment dynamics driven by both natural factors and human activities. This research aims to address knowledge gaps in the application of remote sensing for sediment monitoring by evaluating the effectiveness of three indices: the Normalized Difference Suspended Sediment Index (NDSSI), the Normalized Suspended Material Index (NSMI), and the Normalized Difference Turbidity Index (NDTI) in estimating Suspended Sediment Concentration (SSC) in the section of the Mekong River flowing through Thailand. Landsat-8 satellite imagery, validated with field data collected from 2017 to 2023, was utilized in this study. The findings identified NDSSI as the most accurate index, with an R² of 0.723 and an RMSE of 20.2 mg/L. Comparatively, NSMI showed moderate performance (R² = 0.512, RMSE = 25.8 mg/L), while NDTI exhibited the lowest accuracy (R² = 0.418, RMSE = 27.5 mg/L). The study indicates that NDSSI is the most suitable tool for sediment monitoring in highly turbid river systems, whereas NSMI and NDTI require further refinement to enhance their applicability in complex hydrological environments. This research highlights the significant potential of remote sensing for sustainable sediment management, offering actionable insights to improve monitoring methods and supporting future work in integrating advanced modeling techniques with high-resolution satellite data to effectively address sediment-related challenges.

Transforming Water Management: The Impact of Harris Hawks Optimization on Ubolratana Dam, Thailand

Rapeepat Techarungruengsakul — 2025-06-30

Effective reservoir management is critical for addressing water scarcity and ensuring water security, especially in drought-prone regions. However, traditional reservoir operation methods, such as the Standard Operating Procedure (SOP), often fail to adequately balance water deficits and surpluses under changing climatic and demand conditions. This study addresses these limitations by integrating the Harris Hawks Optimization (HHO) algorithm with a reservoir simulation model, aiming to enhance operational efficiency at Ubolratana Dam in northeastern Thailand. The research evaluates the Hedging Rule (HR) against SOP benchmarks, highlighting its ability to reduce average water shortages and excessive water releases. Using historical management data, monthly inflow patterns, and current water demand, the proposed HR framework demonstrates a 53% reduction in water shortages and a 19% decrease in excessive releases compared to existing practices. These results underscore the significant potential of optimization-based approaches in improving reservoir resilience and reliability. This study fills a critical gap in sustainable water management by offering a robust and adaptable framework for optimizing reservoir operations in regions vulnerable to climate variability.

Exploring Interoperability Factors Influencing Personal Electric Vehicle Adoption in Chiang Mai, Thailand

Krisana Yindee — 2025-06-30

The transportation industry stands on the brink of a transformative revolution as the number of electric vehicles (EVs) continues to rise. However, the potential limiting factor of charging station availability looms large. This research, conducted in Chiang Mai province in northern Thailand, delved into the plans and decision-making processes of stakeholders in this transportation revolution, as well as the factors influencing the choice of EVs by users and potential users. The data, collected from 200 EV users and 200 potential buyers, was analyzed using structural equation modeling analysis. The findings underscore the need for cooperation among government agencies, private sector organizations, and electricity-producing entities to propel Chiang Mai towards a smart city, with a focus on pollution-free EV travel. Environmental concerns and fuel cost savings emerged as key factors influencing the decision to purchase an EV. The choice of an EV was also influenced by brand quality and standards. However, the most significant factor affecting the potential purchase decision was the availability of electric charging stations. To effectively support the use of EVs and meet environmental pollution standards, it is imperative that governmental, municipality, and private organizations concentrate on developing a sufficient power supply and network of charging stations.

Robust Optimization for PV and BESS Configurations in Distribution Network with EV Load Uncertainties

Saksit Deeum — 2025-06-30

The integration of electric vehicles (EVs) into electrical distribution systems introduces significant challenges and opportunities for optimization, particularly in the context of incorporating renewable energy sources such as photovoltaic (PV) systems and battery energy storage systems (BESS). This paper presents a comprehensive study on optimizing PV and BESS configurations within the IEEE 33 Bus distribution system, focusing on addressing the uncertainties of varying EV loads. Advanced optimization techniques such as particle swarm optimization (PSO), genetic algorithms (GA), and quantum-inspired evolutionary algorithms (QEA) are employed to determine optimal sizes and placements for PV and BESS installations. Monte Carlo simulation models EV load variability, ensuring that the optimization framework accounts for real-time data and forecasted demand. Results demonstrate that incorporating EV load uncertainty into optimization significantly enhances system resilience, efficiency, and cost-effectiveness. This research provides valuable insights for utilities and system operators, offering guidance on deploying renewable energy resources and storage solutions to build a more reliable and sustainable energy infrastructure.

Real-Time Fall Detection for Elderly Care Using YOLOv8 with a Custom-Built Image Dataset

Supakorn Ukampan — 2025-06-30

This paper aims to develop a model for human fall detection by simulating authentic fall incidents for implementation in a computer vision system designed to monitor falls in the elderly and deliver real-time notifications. The model development process commences with the utilization of a dataset comprising item bounding boxes and corresponding annotations. The YOLOv8 methodology is subsequently employed to train the dataset. The study dataset consists of 2,788 raw images that have been annotated and processed using Roboflow technology. The images are categorized into three groups: the training set comprises 77% of the data, totaling approximately 2,146 images; the validation set constitutes 12%, or about 338 images; and the test set accounts for 11%, roughly 304 images. Data augmentation methods were used in the fourth stage of the Roboflow platform to increase data diversity, resulting in 19,000 images. This expanded dataset enhances the model's ability to generalize by exposing it to a wider variety of scenarios and conditions. Consequently, the increased volume of images allows for more robust training, ultimately improving the accuracy and reliability of the model's predictions in real-world applications. The ideal value for improving model performance is one hundred epochs, which is how long model training was run. The model testing outcomes, carried out in the same setting as the training, show a mean average accuracy (mAP) of 90.97% and an overall accuracy of 95.36%, suggesting outstanding accuracy and appropriateness for practical use.

Development of PV/Battery Grid-Connected System for Efficient Charging of Plug-in Electric Vehicles in Residential Areas in Thailand

Chanasith Jan-ngurn — 2025-06-30

This paper explores the development of photovoltaic (PV) and battery grid-connected systems to enhance plug-in electric vehicle (EV) charging in residential areas. The study employs a rooftop PV grid-connected system designed for a residential area with typical monthly electricity consumption, aiming to analyze the impact of power usage on household EV charging. The proposed system evaluates and compares four EV brands with power ratings of 105, 110, 120, and 110 kW. Findings indicate that the average daily power demand is 8 kWh, covering both quick and conventional charging modes. Among the four brands, the second brand exhibited the highest energy consumption for fast charging, totaling 3,096 kWh annually, while the fourth brand consumed the most power for normal charging, reaching 2,859 kWh per year. The annual electricity procurement cost from the grid was calculated at 76.80 kWh. PVsyst analysis revealed a system performance ratio of 76.6%, capable of generating 21,826 kWh/year, with 5,390 kWh/year consumed by AC loads. Financial assessments identified the optimal choice for the 120-kW fast charger from the fourth brand. The study confirms that integrating PV with a battery energy storage system offers an efficient solution for improving plug-in EV charging in residential microgrids.

Integrated Control of DER Placement and Network Reconfiguration in EV-Charging Distribution Systems Using Multi-Optimization Techniques

Paramet Nuamkoksung — 2025-06-30

The integration of distributed energy resources (DERs), such as photovoltaic (PV) systems, into power distribution networks is critical for enhancing grid reliability, reducing power losses, and promoting renewable energy adoption. Fast charging stations (FCSs), due to their high energy demand, further complicate grid operation, particularly in maintaining voltage stability and coordinating power supply. While previous studies often address DERs placement or control strategies in isolation, this study proposes a unified framework that optimizes both the placement and sizing of DERs in combination with advanced grid control mechanisms. The proposed approach uses a hybrid of three metaheuristic algorithms: Grey Wolf Optimization (GWO), Particle Swarm Optimization (PSO), and Whale Optimization Algorithm (WOA). The multi-objective formulation focuses on minimizing power loss and cost, improving voltage profiles, and reducing the L-index. A notable contribution of this work is the integration of Volt/Var and power factor (PF) management into the optimization process, which enables practical grid stabilization under steady-state conditions. The methodology is applied to the IEEE 33-bus distribution network and validated through simulation. Results indicate that the hybrid method performs better than traditional single-algorithm approaches, achieving significant power loss reductions and voltage improvements. These findings provide a practical roadmap for distribution system planning under high DERs and FCS penetration.

Depression Classification with Imbalanced Data Problems: Literature Survey

Artitayaporn Rojarath — 2025-06-30

Depression is an increasingly serious global mental health concern, with the number of affected individuals rising steadily. In Thailand, more than 70% of the working-age population is at risk of developing depressive conditions, as reported by the Thai Depression Center. A significant challenge in depression research is the issue of imbalanced datasets, where the number of depressive cases (minority class) is significantly lower than non-depressive cases (majority class). This imbalance often results in biased classification models that favor the majority class, thereby reducing the accuracy and effectiveness of depression classification. This literature survey addresses critical gaps in the field by focusing on the imbalanced data problem in depression classification. While previous studies have primarily relied on traditional oversampling and undersampling techniques, these approaches often intensify the problem of overfitting and lead to the loss of valuable information. Our research explores these issues by reviewing various resampling methods, with a particular emphasis on advanced oversampling techniques that aim to preserve data integrity while mitigating overfitting. The survey also presents a comparative analysis of evaluation metrics, including accuracy, precision, recall, F1-score, and AUC, to provide a more nuanced understanding of classifier performance in the context of imbalanced data. Our findings indicate that while oversampling methods are generally effective, careful implementation is essential to avoid overfitting, which can distort the predictive accuracy of the model.

Unveiling Insights: A Comprehensive Bibliometric Analysis of Generative Artificial Intelligence

G. Lithesh — 2025-06-30

Generative artificial intelligence (GAI) has become prominent in recent days. It has changed the facets of artificial intelligence and is widely implemented in various fields. GAI and its applications have a great influence on society. Hence, to understand its importance and influence well, a comprehensive bibliometric analysis of GAI is proposed in this paper. This bibliometric analysis aims to explore the bibliometric data in terms of challenges, proposed methods, applications, and insights. Further, it is a quantitative tool for evaluating scholarly publications. The proposed bibliometric analysis is performed on the bibliometric data collected from the Scopus (753 records) and Web of Science (400 records) databases ranging from 2013 to 2024 and 448 unique records are considered for the analysis. Further, after scrutiny of these records, 46 records are considered to discuss various applications of GAI. The proposed review is executed systematically by using the PRISMA model. To conduct the analysis, ten critical research questions are identified, and the answers are obtained through the results of the proposed analysis. The key results of this bibliometric analysis unveil various insights into GAI research in terms of impactful applications (22), patterns, research trends, the progress of GAI over the years, scholarly articles production, trending topics, acknowledged collaborative dynamics of authors, affiliations, and countries (10), top influencing authors (10), affiliations (10), and sources (10). These insights drive future aspiring researchers to understand the significance of GAI in various applications and enable them to carry out fruitful research.