Article

• Effects of design parameters on the durability of a shallot-harvesting mechanism

  Nghĩa Nguyễn Hữu, Danh Trần Hữuu, Đăng Nguyễn Thuận hải (Author)

  Abstract | PDF (Vietnamese) | Views: 101 Downloads: 8

  A model of the shallot harvesting mechanism with three blade profiles (scoop-type, trapezoidal, and triangular) was built in Autodesk Inventor and imported into ANSYS Workbench. Using the finite element method, realistic operating loads were simulated to analyze equivalent (von Mises) stress, natural frequencies and harmonic response, and the blade’s vibration amplitude. Based on the simulation/experimental dataset, a Taguchi L27 design was created for four design variables—blade thickness , tip radius , blade spacing , and profile . Second-order response surface (RSM) models were developed and validated with ANOVA and , and 3D contour plots were generated to visualize parameter effects. Multi-objective optimization with a desirability function was performed to reduce stress and vibration amplitude while increasing natural frequency. An optimal set of design parameters was identified, improving durability, reducing vibration, and enhancing the operational stability of the shallot harvesting mechanism.

• Research and manufacture self-driving vehicles using image processing

  Khải Đỗ Huỳnh Quang, Vạn Lương Văn (Author)

  Abstract | PDF (Vietnamese) | Views: 60 Downloads: 9

  The paper presents the process of researching and manufacturing a self-driving car model that uses image processing to recognize lanes, signs, and traffic lights to control the car's movement automatically. The system is built on the Raspberry Pi 4 and Nucleo platforms, combining algorithms such as Canny, Hough Transform and neural networks to process real-time image data. The model is designed to target Vietnam's traffic conditions, with the ability to recognize signs and stop and turn according to real signals. The results show stability and accurate handling in changing light and environmental conditions, contributing to the basis for large-scale autonomous vehicle development in the future.

• Isolating and selecting acetic and lactic acid bacteria from Kombucha

  Kiệt Nguyễn Trần Tuấn, Chi Nguyễn Trí Yến, Nho Trần Quang (Author)

  Abstract | PDF (Vietnamese) | Views: 85 Downloads: 17

  This study aimed to isolate and evaluate the acid-producing abilities of acetic and lactic acid bacteria strains from Kombucha for selecting strains suitable for fermentation. From 5 Kombucha samples, 22 strains were isolated on the YPGD medium and 7 strains on the MRS medium. Morphological and biochemical analysis indicated that 11 YPGD-isolated strains were similar to acetic acid bacteria, with strain A2-5 showing the highest acid production at 24, 30, and 36 hours. The strains isolated on the MRS medium displayed characteristics that differed from beneficial lactic acid bacteria. 16S rRNA gene sequence analysis showed that strain A2-5 is 100% identical to the acetic acid bacterium Komagataeibacter saccharivorans

• The impacts of social networks on students in the digital age

  Ngân Đinh Trần Kim, Trương Bùi Nguyễn Hoàng (Author)

  Abstract | PDF (Vietnamese) | Views: 94 Downloads: 14

  In the 4.0 era, social networks have become popular and brought many practical benefits to people in general, especially students. However, the use of social networks also carries many potential risks and consequences, particularly for young people. This article focuses on analyzing the impacts of social networks on students in the digital age and at the same time proposes solutions to help them use these platforms in a responsible and safe manner, in order to maximize benefits and minimize risks in today's rapidly evolving technological landscape.

• Effect of steaming time on potato flour quality and application to noodle processing

  Trâm Giáp Phạm Ngọc, Yến Trần Tiểu (Author)

  Abstract | PDF (Vietnamese) | Views: 234 Downloads: 17

  The objective of this study was to determine the appropriate steaming time to improve the sensory quality, water absorption capacity, color, and flavor of potato flour (Solanum tuberosum), as well as to investigate the effect of different supplementation levels of potato flour on the quality attributes of noodle products. The evaluated parameters included color values (L*, a*, and b*), cutting force, and sensory characteristics. The analytical results indicated that steaming potato flour for 6 minutes at 90°C yielded the best quality parameters, including color values of L* (73.58), a* (5.52), and b* (29.43); moisture content (7.66%); and water absorption capacity (6.14 g/g). Moreover, the addition of potato flour at a 10% level resulted in optimal noodle quality, as evidenced by color values L* (72.01), a* (4.74), and b* (32.68); cutting force (23.47 N); and high sensory scores in terms of color (8.40), aroma (8.20), taste (8.00), texture (8.13), and overall acceptability (8.30). Steaming potatoes for 6 minutes and using a 10% potato starch blend ratio are suitable specifications for producing noodles with good sensory quality and structural value.

• Identification of cyber attack methods using machine learning techniques

  Khang Trương Thành, Cang Phan Anh, Cang Phan Thượng (Author)

  Abstract | PDF (Vietnamese) | Views: 93 Downloads: 16

  This study focuses on the problem of network attack detection and classification using modern machine learning techniques, including CNN-LSTM, RNN, and LSTM, on the NFUQ-NIDS-v2 dataset. Two experimental scenarios were conducted: (1) training directly on the original imbalanced dataset without any balancing technique, and (2) applying class-weight to address class imbalance. The experimental results indicate that in scenario (1), the models achieved high accuracy for majority classes but failed to recognize rare attack types. In contrast, scenario (2) showed that the class-weight approach significantly improved the detection and classification of all 21 classes, especially minority classes, although the accuracy of some majority classes decreased. Overall, the class weight technique proved effective in mitigating the impact of data imbalance, suggesting future directions to integrate advanced methods such as SMOTE variants to further enhance accuracy and generalization capability in practical deployment. 

• Developing standards for digital learning materials in practical instruction: the case of the restaurant ordering process

  Trương Bùi Nguyễn Hoàng, Long Mai Hoàng (Author)

  Abstract | PDF (Vietnamese) | Views: 52 Downloads: 7

  In the context of digital transformation in higher education and vocational training, digital learning materials play an increasingly important role in practice-based courses; however, the lack of unified standards for developing digital learning materials for practical instruction has limited their pedagogical effectiveness. This study aims to develop a standardized framework for digital learning materials in practical teaching, based on digital education policies, theoretical foundations of digital learning resources, and the specific requirements of skill-oriented lessons, with the restaurant ordering process selected as a case study. The proposed framework includes standards for classifying practical digital learning materials as well as technical, pedagogical, management, and legal requirements, thereby supporting the standardization and practical application of digital learning materials in teaching practice.

• GI-ENSNet: A Hybrid Model For Segmenting Some Gastrointestinal Organs

  Ho Minh Quang, Phan Anh Cang (Author)

  Abstract | PDF | Views: 44 Downloads: 4

  Gastrointestinal cancer is a leading cause of mortality, requiring highly
  accurate treatment approaches such as radiotherapy. Accurate localization of organs such as the stomach, small intestine and large intestine on MRI scans is crucial to minimize radiation-induced damage to surrounding healthy tissues. However, manual segmentation is often time-consuming and prone to subjective errors. This study proposes a deep learning-based model named GI-ENSNet, which ensembles three architectures: UNet with EfficientNet-B2 as backbone, UNet++ with EfficientNet-B0 optimized using Stochastic Weight Averaging (SWA) and a modified UNet integrated with the Boundary Awareness Module (BAM). The outputs of these models are fused using a Soft Voting ensemble strategy, followed by post-processing with the Segment Anything Model (SAM) to refine segmentation masks. The model is trained on a dataset of 38,496 MRI images provided by the UW-Madison Cancer Center and achieves high segmentation performance with a Dice coefficient of 0.9255 and an IoU of 0.8964. The proposed approach significantly enhances segmentation accuracy and supports clinicians in delivering more precise radiotherapy while reducing processing time in clinical practice. 

• NEUMF–LSTM: Improving Target Behavior Prediction in Multi-Behavior Recommender Systems

  Le Thi Hanh Hien, Tran Thu Mai, Nguyen Thai Nghe, Tran Thi To Loan (Author)

  Abstract | PDF | Views: 49 Downloads: 36

  This study proposes a NeuMF–LSTM model for multi-behavior recommender systems in e-commerce, integrating the nonlinear interaction learning capability of NeuMF with the sequential modeling strength of LSTM. Experiments on the Tianchi and Tmall datasets show that the proposed model improves purchase prediction accuracy compared to baseline methods, confirming its effectiveness.

• Electricity Consumption Forecasting for Seafood Processing Plants Using a Hybrid ANN–PSO Model

  Tran Van Si, Nguyen Tan No, Nguyen Tran Quang Huy, Do Chi Phi (Author)

  Abstract | PDF | Views: 53 Downloads: 27

  The study develops a hybrid ANN–PSO model to forecast the electrical load of a seafood processing plant in Sóc Trăng, where consumption fluctuates strongly due to seasonal production and refrigeration systems. Data were collected from an MFM-384-C meter (1 hour/sample for 1 month) and processed through an HMI WEINTEK MT8071iE on an IoT platform. The ANN–PSO model uses a 1–5–1 structure, with PSO optimizing weights and biases to improve accuracy. Results show very low errors: hourly forecasts are mostly <0.5% (hour 1: actual 315 kWh, predicted 314.52 kWh – 0.15%, maximum 1.20%); weekly forecasts mostly <1% (hour 1: 2141 kWh vs. 2138.54 kWh – 0.11%, maximum 4.29%); monthly errors remain stable, with day 19 at 7524 kWh vs. 7525.41 kWh – 0.02%, and the highest error 1.45% on day 30. Compared to traditional ANN, ANN–PSO performs better, reducing RMSE from 1.45% to 0.53% (hourly) and from 1.85% to 0.78% (monthly). The model proves feasible for accurate energy consumption forecasting and energy management in processing plants.

• Quadratic Response Surface Modeling of Surface Roughness in CBN Grinding of HARDOX 500

  Vo Thanh Tuan, Le Hoang Anh, Ho Xuan Tu, Nguyen Manh Cuong (Author)

  Abstract | PDF | Views: 60 Downloads: 17

  This study investigates the influence of grinding parameters on surface roughness (Ra) in grinding HARDOX 500 steel with a CBN wheel, using a quadratic Response Surface Methodology (RSM). A full quadratic model, including linear, interaction, and squared terms, was initially developed and subsequently simplified through stepwise selection based on the Akaike Information Criterion (AIC). The full model achieved R² = 0.987 and adjusted R² = 0.964, while the simplified model achieved R² = 0.985 and adjusted R² = 0.974, indicating strong predictive capability with reduced complexity. Analysis of variance (ANOVA) and partial effect size (η²) revealed that the squared term of depth of cut (t²) exerted the dominant and statistically significant influence on Ra (p < 0.001). In contrast, spindle speed, feed rate, and their higher-order terms played only minor or negligible roles. Diagnostic plots confirmed model adequacy, with residuals evenly distributed and predicted values closely matching experimental data. The findings highlight that precise control of depth of cut is the key factor in achieving the desired surface quality when grinding HARDOX 500 steel, providing practical guidance for optimizing process parameters in industrial applications.

• Research on Effects of Additional Brown’s Gas on The Working Emission Characteristics of Diesel Engine

  Bui Nguyen Hao, Nguyen Thanh Tung (Author)

  Abstract | PDF | Views: 45 Downloads: 5

  The paper presents simulation results aimed at determining the effects of Brown’s gas addition into the intake manifold of a diesel engine on operating characteristics and exhaust emissions along the external characteristic curve. The investigated engine is the IVECO 8140.43R, and the simulation tool employed is AVL Boost. The results indicate that with a Brown’s gas substitution rate of 10÷20% of the torque, the average torque increases by 17%, power output increases by 25%, brake specific fuel consumption decreases by 20%, NOx emissions increase by 11,3% on average, soot emissions decrease by 13,7%, and CO emissions decrease by 25% on average.

• Performance Evaluation of Energy-Saving Solutions For a Thermal Oil-Based Drying System with a Capacity of 1000 Kg/Batch

  Tran Tan Loc, Tran Huu Danh, Do Chi Phi (Author)

  Abstract | PDF | Views: 72 Downloads: 8

  Energy losses in sliced fruit-drying systems employing thermal-oil heaters constitute a critical technical challenge, with measurable impacts on both operational efficiency and environmental performance. This study provides a quantitative assessment of three principal mitigation strategies: 1. Recovery of waste heat from flue gas via installation of an air preheater (APH); 2. Enhancement of insulation and thermal protection; 3.Optimization of the combustion process. When these feasible measures (Solutions 1, 2, and 3) are implemented concurrently, the overall thermal efficiency of the heater can be improved from 87.84% to 95.27%. Building on these results, This study investigates outlines an integrated implementation pathway aimed at maximizing energy utilization efficiency and strengthening the economic performance of fruit-drying operations.