Journal of Engineering and Digital Technology (JEDT)

Real-Time Classification of Optical Devices Using Rotating Linearly Polarized Light in a Sagnac Interferometer

Rapeepan Kaewon — 2026-04-27

This study looks at how to create perfectly rotating linearly polarized light using phase-shifting methods in a Sagnac interferometer, aiming to categorize optical devices. Theoretical analysis is conducted using Jones calculus, which provides a framework for understanding the propagation and phase shifting of linear light. Experimental results from the Sagnac interferometer show interference fringes that align with predictions from mathematical simulations. The experimental observations are validated through comparisons with Python-based simulations, ensuring the accuracy of the rotating polarized light characteristics. Additionally, Convolutional Neural Network (CNN) techniques are employed to analyze and verify the interference fringes, further confirming the consistency of the results with Jones calculus theory. This work demonstrates the potential for applying these methods in real-time, non-destructive optical measurements for the inspection and classification of materials such as polarizers and Half Wave Plates (HWPs), advancing the field of optical device characterization.

The Construction of a Newsvendor Model Based on Conditional Value at Risk (CVaR) and the Determination of Optimal Order Quantity

Meijia Wang — 2026-04-27

To address the growing uncertainty in demand for electric vehicles in the market, this article develops an extended CVaR Newsvendor model. Compared to existing Newsvendors, there are two contributions: (i) embedding both risk aversion and loss aversion in utility-oriented objectives; (ii) handling limited additions using the “three scenarios” (no stockouts/no additions, no stockouts/limited additions, and stockouts/limited additions). Based on 12 months of data from a single retailer, the model was calibrated under a normal distribution demand and fixed cost structure, and evaluated using a Monte Carlo simulation. The model had a high fitting degree (98.08%) and a low root mean square absolute error (0.8 vehicles). This study assumes a single cycle and a single project, with parameters considered stable during the sample period, and does not explicitly incorporate exogenous factors such as policies and input prices. These assumptions limit the research's generality, but clarify how to adjust the framework in practice through scenario-based parameter tuning.

Bucket Wheel Excavator Teeth Reinforcement via Design of Experiments

Sarayut Malaipurn — 2026-04-27

The reinforcement of materials for Bucket Wheel Excavator (BWE) teeth was investigated using the Design of Experiment (DOE) technique. The primary objectives were to identify the causes of wear, determine the material factors affecting wear, and compare the Performance of reinforced teeth across three material types: SS400, S50C, and SCM440. Forty-eight teeth of each type were used on one rotating wheel, sized 38×38×250 mm, with the reinforced teeth weighing no more than 50 kg each.

The DOE analysis revealed that material type significantly impacts hardness (p-value = 0.016) with a very large effect size (η² ≈ 93.63% and high Cohen's d for all Pairs). Specifically, SCM440 achieved the highest hardness and significantly differed from SS400 both statistically and practically. S50C was also practically harder than SS400. While post-hoc tests (e.g., Bonferroni) did not detect statistical differences for all Pairs (e.g., SCM440 vs. S50C), the consistently high Effect Size confirms the Practical significance of these differences. Therefore, selecting SCM440 is the most suitable choice for maximum hardness. This superior mechanical property is consistent with the field experiment’s wear Performance, even under varying abrasive conditions. The field test showed that the high-hardness SCM440 was able to operate in the most severely abrasive environment (100% sand content), which strongly supports the conclusion that SCM440 offers better wear resistance than S50C and SS400, especially in highly corrosive environments.

Applying Discriminant Analysis for Data-Driven Decision Making to Reduce Defects in Integrated Circuit Pick-Up from Dicing Tape: A Case Study of an Electronics Company

Jiranan Jaioer — 2026-04-27

An electronics manufacturer experienced a high failure rate in picking up integrated circuits (ICs) from dicing tape at 43.5 percent, leading to increased waste and production costs. This research aimed to investigate the factors affecting the IC pick-up process by applying discriminant analysis to classify the workpieces into two groups: good and defective. The study also sought to identify the optimal values of key variables contributing to defect reduction in the inspection and packaging processes. A total of 180 actual production data sets were used to build and validate the model. Predictor variables included needle distance (X₁), vacuum suction force of the pick-up head (X₂), and vacuum suction force of the dicing tape needle (X₃). The analysis revealed that the two key factors—needle distance and vacuum suction force of the pick-up head—significantly influenced the success of the process. The model achieved 95.14 percent accuracy in training and 97.22 percent in testing. The optimal settings were 502 micrometers for X₁ and -0.55 millibar for X₂. Additionally, the suction force of the dicing tape needle (X₃) was recommended to be set at -0.15 millibar. This resulted in an average proportion of IC pick-up failures decreased by 13.25 percent and the average number of IC that failed to pull up decreased by 43.36 percent that leads to a disposal cost reduction by 48.25 percent.

Used Car Price Prediction Using Web Scraping and Machine Learning Models

Bhurisub Dejpipatpracha — 2026-04-27

This study aimed to develop a predictive model for used car prices in Thailand using web scraping techniques and machine learning algorithms. Data were collected from Kaidee.com, One2Car.com, and Chobrod.com, totaling 55,989 records. After data cleaning, 42,823 valid records remained, containing 13 fundamental attributes for model construction. Two experiments were conducted: (1) using only the basic features and (2) incorporating six newly engineered features—car age, annual usage rate, squared mileage, squared engine size, cumulative usage load, and temporal load—to enhance the model’s learning capability. The performance of five models, including XGBoost, Random Forest, LightGBM, CatBoost, and Gradient Boosting, was compared using MAE, RMSE, MAPE, R², and accuracy. The results showed that XGBoost achieved the best prediction performance. With the additional features, the R² value improved from 0.9262 to 0.9419, accuracy increased from 89.42% to 92.38%, and MAPE decreased from 10.58% to 7.62%, indicating that feature engineering significantly enhanced model accuracy. Feature importance analysis revealed that the most influential factors affecting used car prices were fuel type, car type, engine size, brand, and squared engine size. The findings confirm that integrating machine learning with feature engineering substantially improves predictive performance and can serve as a decision-support tool for buyers, sellers, and financial institutions to promote transparency and fairness in Thailand’s used car market.

Network Modeling and Analysis of Cassava Mosaic Disease Transmission in Thailand

Chayanusapat Rattanavarawong — 2026-04-27

This study aims to develop and analyze a mathematical model based on a two-node network structure to investigate the spread of cassava mosaic disease in Thailand. The plant population is divided into four compartments according to infection status, and the connections between two regions are considered through the movement of insect vectors and stem cuttings. The mathematical model is analyzed symbolically using the next generation matrix to calculate the basic reproduction number (), and the stability of equilibrium points is examined via the Gershgorin circle theorem. The results show that the infection and recovery rates have the greatest influence on . The model explains the disease dynamics under both disease-free and endemic conditions, highlighting the risk of transmission between regions through the movement of planting materials. Policy recommendations consistent with the finding include controlling the migration of stem cutting, adopting resistant cultivars, and implementing appropriate field management to enhance the sustainable control of cassava mosaic disease.

Influence of Ground Oyster Shell on Properties of Fly Ash-Based Geopolymer Paste

Worawit Projan — 2026-04-27

Geopolymer paste is one type of environmentally friendly binder material with outstanding mechanical properties and high durability. However, heat curing is necessary to accelerate the geopolymerization. This research aimed to study the influence of ground oyster shells on the properties of fly ash geopolymer paste at a normal ambient temperature. High-calcium fly ash and ground oyster shells were used as the main raw materials in the ratios of 100:0, 75:25, 50:50, 25:75, and 0:100 by weight. The liquid-to-powder material ratio at 0.6 and the sodium silicate to 10 M sodium hydroxide ratio at 1.0 by weight were used. Geopolymer paste properties were tested, including flow value, setting time, compressive strength, and microstructure analysis. The experimental results found that the flow rate and setting time of the geopolymer paste decreased when the quantity of ground oyster shells increased. The 75:25 mixture of fly ash and ground oyster shells achieved the maximum compressive strength of 44.3 MPa at 28 days. Microstructural analysis revealed that calcium leached from the shell reacted with silicates and aluminates to form C-A-S-H alongside the main geopolymer gel, resulting in a denser microstructure and enhanced compressive strength. It was shown that geopolymer paste from fly ash mixed with ground oyster shells has high potential for producing environmentally friendly construction materials, with good compressive strength development under room temperature curing.

Sensitivity Analysis of Partial Joint Ordering Policy in Multi-Location Distribution Systems

Anchalee Supithak — 2026-04-27

This research performs a sensitivity analysis on inventory replenishment for multiple demand locations using a partial joint ordering policy. Specifically, the study examines inventory replenishment for dispersed systems comprising five and ten locations. A performance index compares the partial joint ordering policy against a full joint ordering policy by calculating the ratio of their total inventory costs. The problem is formulated as a binary linear programming optimization model and solved using the Excel Solver to minimize total inventory cost. Sensitivity analysis is conducted to evaluate the impact of variable setup costs and demand levels on the total inventory cost of partial joint ordering. Variable setup costs are analyzed across a range from 2,000 to 40,000. The study considers dispersed location systems with both equal and unequal variable setup costs. Furthermore, demand is classified into two cases: equal value of 10,000 and unequal values between 8,000 to 22,000. The results show that for the distribution system having different demands, when variable setup cost increases, the performance index decreases. For five locations problem and ten locations with the unequal demands and high variable setup cost, the partial joint ordering gives the total inventory cost in average lower than the full joint ordering by 1.69% and 2.75%, respectively. The distribution system with high different dispersed demands, the partial joint ordering has an advantage in term of total inventory cost. On the other hand, the partial joint ordering is not advantageous when the dispersed demands are equal or not much difference.