Indonesian Journal of Electrical Engineering and Informatics (IJEEI)
Not a member yet
776 research outputs found
Sort by
Exploration of Analyte Electrolyticity Using Multi-SRR-Hexagonal DNG Metamaterials and ZnO Thin Films
Full text linkAdvanced engineered metamaterials (MTMs) significantly contribute to modern technological advancements, particularly through hybridization with semiconductor materials like zinc oxide (ZnO), which enhance sensor sensitivity and performance. This study aims to investigate the optical properties of hybrid MTMs and develop a novel sensor medium capable of detecting early electrolytic behaviors of analytes. Utilizing the finite-difference time-domain (FDTD) method, the sensor was designed, characterized, and integrated, featuring a hexagonal multi-cell split ring resonator (SRR) structure coated with a 200-nm ZnO thin film. The geometry of the SRR MTM was optimized using a modified Nicolson-Ross-Weir electromagnetic field function method. Results demonstrate that the MTM exhibits double-negative optical characteristics with a performance index reaching 102. Moreover, the sensor presents dual-band resonance frequencies for reflection and transmission attributed to the combination of the multi-SRR hexagonal design and ZnO coating, with an absorption peak at 8.71 GHz. Testing the sensor in varying electrolytic conditions, such as seawater, revealed a measurable reduction in resonance depth and increased sensitivity, characterized by a frequency shift of 5.25 MHz per 0.7 S/m increment in electrical conductivity. These findings highlight the MTM sensor's potential as an effective tool for enhancing spectrum readout accuracy and sensitivity in analyte detection applications
Exploratory Analysis of the Impact of Data Balancing on the Classifier’s Performance in Predicting Creditworthiness Reliability
Full text linkThis study examines the application of machine learning algorithms for creditworthiness prediction within the banking sector and addresses the issue of class imbalance through sampling methodologies. The research indicates that using the Stacking Ensemble algorithm with random oversampling can predict creditworthiness with an impressive 93% accuracy. The method consistently achieves excellent precision, recall, and F1-score values, indicating that it can produce accurate predictions while maintaining a balanced evaluation. Random oversampling helps models improve their predictive accuracy and reduce class imbalance. The research findings underscore the feasibility of this technique for financial institutions, facilitating informed lending decisions and improving credit risk assessment methodologies. This research enhances the field by identifying the most effective machine learning methods for accurate creditworthiness evaluation. Using XAI tools like Shapash provides financial organizations with valuable insights into assessing loan risks and enhancing their lending operations
Efficient Strategies for a Medium Voltage Loop Powered by an Infinite Source
Full text linkThis paper analyzes and examines the potential of an infinite generation system to support the domestic load growth of the 33 kV loop network from 2025 to the year 2040. The study assesses the current state of the network, focusing on voltage levels, line loadings, and transformer capacities to ensure that all components operate within the system's allowable loading limits. It is assumed that the loop is powered by an infinite source. A numerical model, utilizing the Gauss-Seidel method, is developed and run using the PSS/E simulator and ETAP. The voltage profile is expected to remain within the range of 0.95 to 1.05 pu. An analysis of the simulation results demonstrates the potential for increasing active power transfer and controlling reactive power in the system by the year 2040.Furthermore, solutions are proposed to address identified critical issues in order to meet the projected demand. These include doubling the capacity of existing transformers and implementing protection against short-circuit currents. The proposed system is expected to provide industrial consumers with reduced load imbalances and improved control over voltage fluctuations caused by rapid changes in reactive power demand
Advanced Classification of Agricultural Plant Insects Using Deep Learning and Explainability
Full text linkThis paper investigates the effectiveness of six pre-trained deep learning models to classify images of agricultural plant insects. We utilized the BAUInsectv2 dataset, which includes images from nine classes. Aphids, Armyworm, Beetle, Bollworm, Grasshopper, Mites, Mosquito, Sawfly, and Stem borer. The models, namely Xception, MobileNetV2, ResNet50, EfficientNetV2B3, ResNet101, and DenseNet121, are fine-tuned by transfer learning from ImageNet. This approach significantly reduces training time while improving classification accuracy. Our experiments reveal that each model reliably distinguishes between insect species even when faced with varying lighting conditions and diverse viewpoints. To further clarify how these models make predictions, we employ Gradient-weighted Class Activation Mapping (Grad-CAM) to highlight critical regions in the images. The results demonstrate that each model focuses on unique biological features and offers clear explanations for its decisions. The research results contribute to demonstrating the potential of pre-trained deep learning architectures for agricultural monitoring and pest management, paving the way for promising future applications
IVFD: An Intelligent Video Forgery Detection Framework Leveraging InceptionV3 and GRU for Enhanced Forensics
Full text linkCloud computing-like services that are great at paying for and managing multimedia are fundamental technological innovations that have made it easier for individuals and organizations to adopt multimedia content. Thanks to social media, different people with different perspectives can voice their opinions and present data through photos and videos. However, video tampering is a significant issue because illegal modification of video content can easily mislead audiences and make it difficult for them to relate to reality. This is, therefore, a serious problem, as the consequences of video forgery are dire. Several image processing-based solutions have emerged to address video forgery. Artificial intelligence has recently allowed deep learning models to be trained extensively; hence, deep learning has been frequently used for video tampering detection. However, further work is still required to refine such models or develop hybrid models to improve the existing models' capabilities in identifying video forgeries and assisting digital forensics. We introduce a framework based on deep learning to automate the detection and localization of video forgeries. We offer a hybrid deep learning model that fuses Inception V3 with a Gated Recurrent Unit (GRU) as part of our framework. We also propose a new algorithm, Intelligent Video Forgery Detection (IVFD), to detect the forgeries and their invariants based on this hybrid model. Through empirical studies applied on a standard dataset, called the Deepfake Challenge dataset, we get an accuracy of 97.21%, which makes our hybrid deep learning model outperform many existing models. Since video content is prevalent in almost all applications in today's era, our design system should be laid on top of these applications, which can facilitate detecting the tampering of the videos and thereby contribute towards digital forensics
Fractional Order Sliding Mode Control to Mitigate Power Quality Issues using Dynamic Voltage Restorer in Distribution Network
Full text linkPower quality (PQ) issues lead industrial customers to suffer significant financial losses. These PQ issues are garnering more attention from electricity suppliers and consumers in the modern day. This study addresses prevalent PQ issues, namely voltage sag and swell, stemming from a decrease in RMS voltage within electrical networks, particularly impacting sensitive loads. The solution proposed involves employing a series connected custom power device (CPD) named as dynamic voltage restorer (DVR) with an integrated DC battery for energy storage, to consistently maintain the requisite voltage magnitude. To effectively combat voltage sag and swell, the study introduces a novel control strategy known as fractional order sliding mode control (FOSMC). Noteworthy features of the FOSMC methodology include its capacity to autonomously and dynamically address sag and swell issues. The Simscape toolbox of MATLAB®/Simulink® is used to perform simulations to showcase the efficacy of the FOSMC technique. The results demonstrate that this strategy ensures total harmonic distortion remains below 5% and achieves sag/swell mitigation in less than 2 milliseconds, aligning with SEMI-F-47 and IEEE voltage standard 1159-2019. In summary, the study introduces and validates a robust control strategy implemented in a DVR system to autonomously alleviate voltage sag and swell issues, with simulation results supporting its effectiveness in upholding PQ standards. The FOSMC scheme with DVR is also compared with FOSMC scheme with DSTATCOM as well as with super twisting sliding mode control (STSMC) algorithm and classical sliding mode controller (SMC) to show the effectiveness of the proposed scheme. The FOSMC technique with DVR is more effective in restoring voltage sag/swell and PQ issues
FOC-Based Soft Start of Induction Motors Using Trigonometric S-Curve
Full text linkThis paper presents a novel approach to improving the starting performance of three-phase induction motors by integrating an optimized S-curve acceleration profile based on trigonometric functions into a Field-Oriented Control (FOC) framework. Unlike conventional third- and fifth-order polynomial trajectories that suffer from limited jerk continuity and insufficient mechanical damping, the proposed method ensures smooth transitions in acceleration and jerk using sinusoidal functions. The core contribution of this work lies in the development and application of a second-order continuous trigonometric velocity trajectory that significantly reduces mechanical shocks and current oscillations during motor startup and stop phases. Furthermore, the method is designed for real-time implementation on FPGA hardware, enabling high-resolution pulse-width modulation (PWM) suitable for embedded motion control systems. Simulation and experimental results demonstrate superior motion smoothness, improved torque tracking, and enhanced mechanical reliability compared to traditional methods. This research provides a practical and effective solution for applications requiring precise soft-start/stop capabilities, particularly in elevator systems and other high-performance industrial drives
Supporting Communication for Deaf People with Sign Language Recognition Using Deep Learning Approach
Full text linkSign language recognition (SLR) plays a crucial role in improving communication for deaf individuals. This paper investigates the recognition of sign language through deep learning models based on action features using Skeleton data from the Argentinian Sign Language (LSA64) dataset. The models explored include Multi-layer Perceptron (MLP) Neural Network, and Long Short-Term Memory (LSTM). The MLP Neural Network, utilizing multiple layers of perceptrons, reached an accuracy of 96.10%. The LSTM model, excelling in processing sequential data, attained the highest accuracy at 98.60%. These results demonstrate the effectiveness of deep learning models in sign language recognition, with LSTM showing the most promise due to its ability to effectively capture temporal dynamics. Consequently, this study opens up prospects for applying sign language recognition technology in practice, contributing to enhancing the quality of life for deaf individuals
A Comprehensive Survey on Artificial Intelligence – Based Classification of Gastrointestinal & Oesophageal Cancers
Full text linkThe global incidence of Gastrointestinal (GI) disorders has risen dramatically over recent decades, driven chiefly by changes in dietary patterns and lifestyle behaviours; epidemiological evidence attributes nearly two million deaths annually to these conditions, underscoring their substantial burden on healthcare systems. Despite endoscopy’s status as the diagnostic standard for detecting mucosal lesions—such as adenomatous polyps and oesophagitis— its performance is hindered by observer variability, limited reproducibility, and lengthy procedural times. To address these limitations, computer-aided diagnostic (CAD) frameworks have been integrated into clinical workflows, offering enhanced accuracy, throughput, and operational efficiency. AI-based pipelines leveraging advanced Machine Learning (ML) and Deep Learning (DL) architectures have proven highly effective in the early detection of GI malignancies and in quantitatively assessing tumour invasion depth. These technologies not only accelerate critical clinical decisions but also support the development of individualized, precision oncology regimens. This survey provides an in-depth assessment of current ML and ML methodologies applied to GI and oesophageal cancer diagnostics, evaluates established prognostic biomarkers, compares algorithmic performance metrics, and identifies key research directions to overcome existing methodological and translational challenges. Although AI-driven diagnostic systems hold the potential to transform GI oncology by standardizing workflows and improving detection rates, their routine clinical adoption requires rigorous validation in multicentre trials and the establishment of comprehensive implementation guidelines
Cybersecurity Implications of 5G Networks: Threats, Potential Vulnerabilities, and Their Implications for National Security and Privacy
Full text linkThe rapid expansion of Fifth Generation (5G) networks represents a revolutionary shift in telecommunications technology, offering increased speed, higher connection density, and enhanced network efficiency. Nevertheless, these benefits have also attracted various security risks that threaten the protection of national security and the privacy of private citizens.
This research investigates the cybersecurity challenges associated with 5G networks by analysing emerging threats, assessing vulnerabilities in 5G infrastructure, and evaluating their impact on national security and individual privacy.
The research approach includes a literature review of various sources of knowledge and regulations or policies, as well as a quantitative analysis of network vulnerabilities through penetration testing and threat modelling.
The study's results indicate that network slicing introduces new risks to a network, as it provides potential attackers with easy access to weaknesses that exist within isolated network slices. Furthermore, the incorporation of Internet of Things (IoT) devices increases the overall risk, as they often lack proper security measures. Lastly, the multi-tenant characteristic of 5G networks poses a challenge in creating secure isolation between various operators and service providers. This makes it imperative for organisations and service providers to enhance their security measures, such as encryption and access control policies, as well as overall policies, to help rectify these issues.
These findings concluded significant implications across national security and privacy fronts. The study re-emphasizes the importance of a multi-sectoral approach to cybersecurity by industries, policy-makers, and academic scholars. Measures and techniques that are relevant to implementing specific safety tactics and regulations were proposed. The results of this study serve as a reference for 5G cybersecurity. The results offer recommendations that are useful in developing security measures to counter threats and improve the security posture of future 5G networks