Journal of Applied Science and Technology

DESIGN OF A BRAIN-INSPIRED EMOTIONAL CONTROL SYSTEM FOR NONLINEAR MASTER-SLAVE CHAOTIC SYNCHRONIZATION

2025-09-26T14:16:20+07:00

This paper presents a brain emotional controller (BEC) for a class of nonlinear systems. The control system consists of a BEC and a robust controller. BEC is a mathematical model that approximates the judgment and emotion of a brain. As well as the sensing algorithm, the emotional algorithm allows fast learning for the BEC. The BEC contains a prefrontal cortex and an amygdala that effectively reduces the tracking error and adjusts the learning error quickly.

APPLYING MACHINE LEARNING TO PREDICT STUDENT ATTENDANCE IN VOCATIONAL EDUCATION PROGRAMS FOR LOWER SECONDARY GRADUATES (9+ SYSTEM)

2025-09-26T14:16:23+07:00

Attendance plays a pivotal role in determining academic performance and preventing dropout among students in Vietnam’s 9+ vocational training system. Early identification of students exhibiting low attendance is critically important for deploying effective support interventions. This study employs machine learning algorithms—including Decision Tree, Support Vector Machine (SVM), Random Forest, and Gradient Boosting—to predict attendance levels, based on learning characteristics, behavioral patterns, family background, and extracurricular activity participation. The Decision Tree model achieved optimal performance, with an accuracy of 93% when using an extended feature set. Restructuring the output labels into two classes (High–Low) contributed significantly to enhanced classification efficacy. The findings provide a foundation for implementing early warning systems to improve student management practices in vocational colleges.

A COMPARISON OF DATA BALANCING METHODS FOR MEDICAL DATA

2025-09-26T14:16:30+07:00

Class imbalance is a common issue in medical datasets and affects the performance of deep learning models in medical diagnosis. This study aims to evaluate different data balancing methods for medical imaging classification tasks. Experiments were conducted on two datasets, chest X-rays for pneumonia detection and brain MRI scans for tumor classification. Traditional resampling methods (downsampling and oversampling), advanced oversampling techniques (SMOTE, ADAYSN), and Generative Adversarial Networks (GANs) were implemented and compared. Performance metrics including recall, F1-score, ROCAUC, and G-Mean were used to assess the effectiveness of each method. Our results demonstrate that GAN-based approaches consistently outperform traditional techniques across various evaluation metrics, with notable improvements in F1-score (2.95 - 52.4%), ROC-AUC (7.93 - 19.5%), and geometric mean (6.28 - 36.3%). This study provides valuable insights for researchers and practitioners seeking to improve diagnostic models trained on imbalanced medical datasets.

OPTIMIZING POWER ALLOCATION ALGORITHM FOR DOWNLINK COUPLINING USERS IN NOMA NETWORKS

2025-09-26T14:29:32+07:00

In this paper, we consider the combination of non-orthogonal multiple access (NOMA) and ratesplitting (RS) in multi-antenna scenario. Specifically, a common signal is generated at the base station (BS) and transmitted with the user’s private signal. By using successive interference cancellation (SIC), the common signal is decoded and removed at each user, and then the private signal can be decoded based on the NOMA scheme. By jointly allocating power and pairing users, a weighted total rate maximization problem is formulated with quality of service (QoS) constraints, in which the priority of different signals is expressed. To solve this problem, a pairing based algorithm is analyzed and proposed. The simulation results show that: 1) the proposed RS-NOMA system outperforms the RS and NOMA systems; 2) the power allocation scenario and user pairing algorithm can improve the RS-NOMA system in terms of weighted sum ratio and outage probability.

ENHANCED CRANIOPLASTY IMPLANT DESIGN VIA SURFACE REGISTRATION

2025-09-26T14:38:33+07:00

Cranioplasty requires precise, patient-specific implants to restore skull geometry while satisfying clinical and surgical constraints. Traditional implant design methods, particularly symmetry-based mirroring, leverage the skull’s inherent symmetry to create individualized implants but are limited by assumptions of perfect symmetry with midsagittal plane, restricting their applicability to non-symmetrical skulls. Meanwhile, data-driven deep learning approaches, despite their promise, often produce lowerresolution implants requiring extensive post-processing and may lack personalization for unique anatomical features. This study proposes an enhanced mirroring technique integrated with surface registration to relax the strict symmetry requirement, enabling high-resolution implant design for imperfectly symmetrical skulls. The proposed implant generation workflow involves several key steps: initial Computed Tomography (CT) image acquisition, skull image segmentation, and a preliminary symmetric mirroring of the skull along the x-axis. The core innovation lies in applying surface registration to accurately align the original and mirrored images, determining a precise transformation matrix. This transformation is then used to create the implant by subtracting the original image from the transformed one, followed by final adjustments based on specific medical requirements. Experimental results demonstrate that the generated implants are highly individualized, exhibit high accuracy, meet clinical demands, and offer improved resolution compared to those produced by artificial intelligence (AI)-based methods. By addressing the limitations of both classical and AI-driven approaches, this work expands the utility of mirroring techniques, offering a practical, high-fidelity solution for personalized cranioplasty implants.

APPLICATION OF PLC AND HMI TECHNOLOGIES IN DEVELOPING A PRACTICAL TRAINING DEVICE FOR INVESTIGATING INDUSTRIAL SENSOR PARAMETERS

2025-09-26T14:46:14+07:00

In the context of increasing demands for the quality of engineer training in the field of automatic control, the development of modern-standard practical training devices is essential. This paper presents the design and fabrication process of a practical model that enables students to examine key technical parameters of industrial sensors, including sensing distance, sensing area, and response frequency. The system is built based on the integration of Mitsubishi PLC control technology, an HMI monitoring interface, and a servo motor to precisely control and collect survey data. Experimental results show that the system provides high accuracy and stability, making it well-suited for teaching and research purposes in the Department of Electrical and Electronic Engineering at Hung Yen University of Technology and Education.

RESEARCH ON DESIGN AND SIMULATION OF SMALL ELECTRIC CAR POWER TRANSMISSION MODEL

2025-09-26T14:55:01+07:00

The increasing emphasis on energy conservation and environmental sustainability has accelerated the transition toward electric vehicles as a key component of future transportation systems. Among the critical subsystems, the power transmission system plays a central role in determining the overall efficiency and performance of electric vehicles. This study focuses on the design and simulation of the transmission system for a small electric car, with a particular emphasis on analyzing its dynamic behavior. Additionally, key performance characteristics such as wheel torque are derived based on simulation results obtained from the developed vehicle model.

OPTIMIZING THE PARAMETERS OF VOLTAGE, CURRENT INTENSITY, AND PULSE TIME FOR MATERIAL REMOVAL RATE ON THE EDM MACHINE WHEN MACHINING C45 STEEL

2025-09-26T15:01:20+07:00

This study investigates the influence of voltage U, current intensity I, and pulse duration Ton on the material removal rate MRR in electrical discharge machining (EDM) on C45 steel. The work uses the Taguchi optimization methodology with three input parameters (voltage U, current intensity I, pulse duration T on), and each parameter set at three levels, resulting in a total of 9 experiments to evaluate the MRR for optimal efficiency. The results indicate that the material removal rate reaches the highest efficiency at 0.355 g/min with the most optimal set of parameters (U = 30V, I = 1.5A, Ton = 80μs). This study serves as a foundation for optimizing parameters to achieve the highest material removal rate in EDM processing.

A STUDY AND EVALUATION OF GARMENT PRESSURE MEASUREMENT METHODS TOWARD THE DEVELOPMENT OF SUITABLE MEASURING DEVICES FOR THE TEXTILE INDUSTRY

2025-09-28T21:18:12+07:00

Clothing pressure is a critical factor influencing wearer comfort, functional performance, and physiological compatibility, particularly in tight-fitting garments such as shapewear, medical compression stockings, and sportswear. Accurate measurement of the pressure exerted between clothing and the human body is essential for optimizing product design and ensuring objective quality control.

This study provides a systematic overview and evaluation of current garment pressure measurement methods, which are categorized into two main groups: direct and indirect measurements. Each method is analyzed in terms of its operating principles, measurement accuracy, applicability, and suitability for different stages of the garment design and development process.

The results indicate that direct measurement methods—especially those using pneumatic pressure sensors—offer high accuracy, minimal interference with the wearer, and strong potential for advanced research and product validation. In contrast, indirect and simulation-based methods are more suitable during the early design phase due to their flexibility and rapid feedback capabilities. Each type of pressure measuring device is appropriate for specific garment types, depending on factors such as contact surface requirements, measurement range, and testing conditions.

Therefore, the development of next-generation garment pressure measurement systems—with capabilities for real-time, multi-point monitoring and integration with virtual body models—is essential. Such systems are expected to enhance design capabilities and contribute to improved comfort and usability for end users.

RESEARCH ON THE MECHANICAL PROPERTIES OF SINGLE JERSEY KNITTED OF RECYCLED NYLON AND ELASTANE YARNS

2025-09-28T21:30:27+07:00

Single jersey knitted fabrics are widely used in clothing due to their good properties of durability, stretch, and air permeability. Knitted fabrics with elastane yarns will have improved elasticity, dimensional stability, and are suitable for sportswear, swimwear, etc. This paper, examines the physical and mechanical properties of single jersey fabrics knitted of Nylon and recycled Nylon yarns with elastane. Two single jersey fabric samples knitted of Nylon and recycled Nylon fibers yarns with elastane yarns are N610 (77% Nylon - 23% Elastane) and N813 (87% Recycled Nylon - 13% Elastane). In particular, the main raw material in the study is recycled nylon fabric, which is the trend of developing raw materials for the textile industry in the future to reduce environmental pollution from plastic waste. The results showed that recycled nylon fabric with elastane yarns has the following properties: Strength, elongation, residual stretch and good breathability, suitable for sportswear fabrics following the green production trend of Vietnam's textile industry.

MOLECULAR DYNAMICS STUDY OF ANISOTROPIC MECHANICAL RESPONSE IN SINGLE-CRYSTAL D0₂₂-TiAl₃ UNDER UNIAXIAL COMPRESSION

2025-09-28T21:49:33+07:00

This study investigates the anisotropic mechanical properties of single-crystal D0₂₂-TiAl₃ under uniaxial compressive loading through molecular dynamics (MD) simulations. The D0₂₂-TiAl₃ alloy, characterized by its body-centered tetragonal (BCT) structure with lattice constants a = 3.850 Å and c = 8.584 Å, was subjected to compressive stresses along the crystallographic a-axis and c-axis. The results reveal significant anisotropy in the mechanical response, with distinct differences in stress-strain behavior, phase transformation, dislocation evolution, and shear strain distribution between the two loading directions. Compression along the c-axis exhibited a lower critical strain and higher peak stress compared to the a-axis, indicating earlier structural failure and higher load-bearing capacity. The findings underscorethe directional dependence of deformation mechanisms in D0₂₂-TiAl₃, which is critical for optimizing the material’s performance under specific loading conditions.