Engineering Journal (Faculty of Engineering, Chulalongkorn University, Bangkok)
Not a member yet
1223 research outputs found
Sort by
A Comprehensive Review of Home Health Care Routing and Scheduling Optimization
Full text linkHome Health Care (HHC) delivers medical services to patients’ homes, supporting recovery, maintaining health, and reducing hospitalizations. The Home Health Care Routing and Scheduling Problem (HHCRSP) addresses the design of caregiver schedules and patient visit routes. This review analyzes studies from 2006 to mid-2024, providing a structured synthesis of HHCRSP research based on problem types, objectives, constraints, benchmark instances, and solution methods. Problem types are classified by input data characteristics as deterministic, dynamic, and stochastic, with increasing attention to dynamic and stochastic cases that better capture real-world uncertainty. Objectives are grouped by stakeholder perspective--organization, caregiver, and patient--highlighting trade-offs among cost efficiency, workload balance, and patient satisfaction. Constraints are categorized into assignment, temporal, and geographic types, with caregiver qualifications and time windows most frequently addressed. A comprehensive synthesis of benchmark instances offers practical guidance for dataset selection and comparison across studies. Solution approaches are dominated by local search and hybrid algorithms, with hybridization gaining prominence since 2016. The review concludes with insights on emerging trends toward uncertainty-aware and stakeholder-integrated HHCRSP models
PDCA Method Application to Mitigate No Hole Drilling Defects in Housing Large RCL Products in the Automotive Sector
Full text linkThis study investigates the application of the PDCA (Plan-Do-Check-Act) method to mitigate drilling defects, specifically No Hole defects, in Housing Large RCL products at HMG, Ltd. The research aims to enhance product quality and reduce customer complaints by systematically identifying root causes and implementing targeted improvements. Analysis using fishbone diagrams identified suboptimal layout design and insufficient operator training as primary contributors to defects. During the Do stage, corrective actions included layout redesign of the inspection area, creation of detailed Work Instructions, and introduction of a Warning Sheet highlighting past quality issues. Evaluation post-implementation showed a measurable reduction in No Hole defects from 21 units (56.8%) to 12 units (42.9%), confirming the effectiveness of changes. The Act stage focused on standardizing improvements through training, updated documentation, and monitoring, ensuring integration into daily operations. This study demonstrates PDCA's efficacy in reducing specific defects and enhancing manufacturing quality, emphasizing continuous improvement toward defect-free production and enhanced customer satisfaction
Integer Linear Programming for Optimizing Drone-Based Delivery Routes
Full text linkThe growing demand for rapid and effcient delivery solutions, especially in healthcare and remote logistics, presents unique challenges for drone routing, including limited battery life, restricted flight zones, and the need for effcient path optimization. This paper addresses these challenges by proposing an Integer Linear Programming (ILP) model to optimize multipoint drone delivery routes. The objective of the model is to minimize total travel distance or time during drone delivery operations while ensuring that each designated delivery location is visited exactly once, with a return to the starting point. The ILP model incorporates practical constraints such as battery limitations, maximum allowable flight distance, and avoidance of no-fly zones, making it suitable for real-world drone delivery applications. Evaluation across four distinct scenarios, including urban and mixed environments, demonstrates that the ILPbased approach enhances route effciency, achieving an average reduction of approximately 25% in total travel distance compared to heuristic methods, specifically the Nearest Neighbor (NN) heuristic. Moreover, it outperforms metaheuristic methods like Genetic Algorithm (GA) and Ant Colony Optimization (ACO) in both distance effciency and adherence to constraints. The ILP model demonstrates computational feasibility, solving problems with up to 50 delivery points in approximately 3 minutes on average. These findings highlight the potential of the ILP framework as a robust tool for optimizing drone delivery networks, offering significant improvements in operational effciency and scalability
Evaluating the Use of Graphite and BeO Moderators for Micro Molten Salt Reactor
Full text linkGraphite is the most common material proposed as the moderator for thermal molten salt reactor (MSR). Apart from graphite, beryllium oxide (BeO) emerges as an alternative candidate of moderator material which offers superior moderating capability that allows more compact reactor core design. This is important especially for microreactors. This study evaluates the use of graphite and BeO as moderator material in micro-sized MSR. The evaluation encompassed neutronic and burnup calculations, using MCNP6.2 radiation transport code and ENDF/B-VII.0 cross section library. Two alternative fuel options, low-enriched uranium (LEU) and LEU combined with thorium (LEU-Th), were analysed. The moderator options were arranged in three different configurations. From the calculation results, the optimal use of graphite and BeO depends on the fuel option. Whilst BeO shows a glimpse of benefit for LEU-Th fuel option, graphite is a more suitable choice for LEU fuel. The use of BeO in LEU fuel failed to offer longer operational time essential in microreactor, but successfully extended operational time in LEU-Th. Inherent safety aspects and the change of fuel materials over time are also discussed
Integrated Solar-Powered System for Water Purification and Green Hydrogen Production: A Statistical Analysis of Performance and Efficiency
Full text linkThis study presents a professionally engineered, integrated solar-powered system designed to simultaneously address water scarcity and advance the clean energy transition. The system combines a solar-driven water purification unit, optimized to meet Omani standards, with a solar-powered electrolysis process operating at 1.9 V, achieving a hydrogen production rate of 152 ml/min. Comprehensive performance assessments were conducted, measuring key parameters including pH, total dissolved solids (TDS), total suspended solids (TSS), biological oxygen demand (BOD), chemical oxygen demand (COD), and hydrogen production efficiency. Statistical analyses using paired t-tests and ANOVA revealed significant improvements in water quality following treatment, with substantial reductions in conductivity (t = 25.34, p = 0.00001), TDS (t = 22.45, p = 0.00002), turbidity (t = 17.56, p = 0.00003), and salinity (F = 5.23, p = 0.02). Furthermore, hydrogen output demonstrated a statistically significant increase with higher voltages (F = 45.67, p = 0.00005). The system incorporates waste heat recovery to enhance overall efficiency. Life cycle assessment and cost-benefit analysis confirm the system’s economic and environmental advantages compared to conventional technologies. This integrated solution supports Oman’s strategic objectives in water security, economic diversification, and renewable energy deployment, while contributing to global clean energy and hydrogen economy initiatives
Comparison of Thai and English Speaking Signals from Brain Using Deep Learning and EEG
No full textThis study investigates the decoding and comparison of brain signals associated with spoken Thai and English words using deep learning techniques and EEG equipment. In the field of Brain-Computer Interfaces (BCI), researchers have extensively explored methods to decode brain signals into text. Two primary approaches exist: invasive (e.g., ECoG) and non-invasive (e.g., EEG). Invasive methods require surgery and offer high-quality signals but carry infection risks. Conversely, non-invasive methods employ scalp electrodes, resulting in lower signal quality but greater practicality for daily use. The present research utilizes three datasets each for Thai and English to evaluate the effectiveness of EEG and compare the outcomes for both languages. The Thai word data consists of three sets: single words (หิว, ปวด, เจ็บ, หนาว, ร้อน), two-word phrases (หิวมาก, ปวดท้อง, เจ็บแขน, หนาวมาก, ร้อนมาก), and three-word sentences (ฉันหิวมาก, ฉันปวดท้อง, ฉันเจ็บแขน, ฉันหนาวมาก, ฉันร้อนมาก). The English word datasets correspond semantically to each Thai set. All results are tested and compared using two machine learning approaches: Multi-Layer Perceptron (MLP) with statistical features and Convolutional Neural Network (CNN) with stacked spectrogram features. The MLP achieved an overall accuracy of 98%, while the CNN achieved 64%
Machine Learning for Water Level Prediction in the Chao Phraya River Basin
No full textHigh precision of hydrological prediction is crucial for real–time operation of flood and drought risk mitigation and strategic planning. This study assessed the predictive performances of three machine learning algorithms; Extreme Gradient Boosting (XGBoost), Random Forest (RF), and Deep Neural Networks (DNNs) for water level prediction. Accordingly, the one–day and one–week water level prediction models for six key gauged stations along the Chao Phraya River and its major tributaries were developed. Selecting input features was carried out based on the physical river–reservoir system using past water level, rainfall, controlled reservoir outflow, and upstream discharges with different travel times. The statistical evaluation indicated that both XGBoost and RF with rainfall input robustly outperformed than DNNs, as it strongly achieved higher R2 from 0.937 to 0.999 for model training and from 0.743 to 0.995 for model testing and lower MAE, MSE, and RMSE values for all daily prediction scenarios. Among these algorithms, RF demonstrated the superior performance for low water level prediction exhibiting the smallest percentage error of overestimating lying between +0.0088% and +0.9380%. XGBoost, RF, and DNNs algorithms exhibited small average percentage errors for high water level prediction ranging from –2.2696% to +1.1587%. Additionally, daily model can capture the entire testing dataset with high precision than weekly model. Daily predictions provide valuable real–time insights for forecasting water levels during critical flood and drought periods. In contrast, weekly predictions assist in strategic water resource planning to address challenges in diverse hydrological environments
Performance Evaluation of HD Map-Aided Localization for Land Vehicles in Underground Parking Using Integrated INS/GNSS and Low-Cost LiDAR
No full textThere are many algorithms for localization and navigation, but achieving accurate positioning can be challenging, especially in heavy traffic or an unfriendly environment, and with low-cost, precise systems. Maintaining reliable positioning in these situations becomes difficult when GNSS signals are lost. To improve autonomous driving (AD), we need to advance research on vehicle localization techniques. This study presents a method that combines INS, GNSS, HD Maps, and Low-Cost LiDAR using the Extended Kalman Filter (EKF). We aim to enhance the performance of localization and navigation for AD by keeping positioning errors acceptable and supporting real-time decision-making in real-world scenarios. We tested this method in a challenging environment without GNSS signals, specifically an underground parking facility beneath the National Cheng Kung University (NCKU) library, where GNSS signals were blocked entirely. Our navigation algorithm achieved impressive 3D positioning differences of less than 0.30 m. The integration system showed significant performance improvements compared to the TC-INS/GNSS system. Compared to the traditional LC-INS/GNSS system, our approach improved by about 94.66%, 90.42%, and 93.93% in the ENU components. This accuracy is essential for precise “where-in-lane” positioning in AD applications
Multi-Objective Optimization of Surface Roughness and Material Removal Rate in Turning SKD11 Steel: A Combination of GP and MOPSO Algorithms
No full textThis study investigates the optimization of surface roughness (Ra) and material removal rate (MRR) through experimental analysis and a combination of Genetic Programming (GP) and Multi-objective Particle Swarm Optimization (MOPSO) algorithms. A total of 65 experiments were conducted with different cutting parameters, including cutting speed (V), feed rate (f) and depth of cut (d). Analysis of Variance (ANOVA) results indicated that f significantly affects Ra, contributing to 87.46% of the total variance, while d and V had a lesser impact. For MRR, all three parameters showed significant effects, with d contributing 52.75% of the total variance. A predictive model was developed by the GP algorithm showed high accuracy with an R² of 0.978 for the training set and 0.934 for the validation set, demonstrating the model's reliability in predicting Ra values. The Pareto-optimal solutions of MOPSO showed stable convergence and identified a broad range of feasible solutions, with Ra values between 0.463 μm and 3.748 μm and MRR ranging from 213.723 mm³/min to 641.250 mm³/min. Validation experiments confirmed the accuracy of the optimization, with deviations between predicted and actual values of less than 8.43% for Ra and 0.05% for MRR. These results indicate that the proposed GP and MOPSO combination effectively optimizes machining processes, providing critical information about MRR and Ra in SKD11 steel turning
A Novel Subset Graph Algorithm for Generating Reversible One-Dimensional Cellular Automata
No full textIn this study, a novel algorithm for generating reversible rules with null boundary conditions for one-dimensional Cellular Automata (CA) is presented. The neighborhood vector of the CA is used by the procedure to create a subset graph. It finds reversible transition rules by examining the connectivity attributes of the graph itself. By ensuring a distinct predecessor and successor for every configuration, this assures bijectivity. In fields like complex system simulations and cryptography, reversibility is essential. This method overcomes the drawbacks of previous approaches, such as the complexity of de Bruijn graphs and the scalability issues with transition matrices. The suggested method's scalability and usefulness are demonstrated by theoretical analysis and illustrative examples. The results suggest the algorithm's efficiency in generating reversible CA rules, making it suitable for various applications requiring precise and reliable computational reversibility