Bulletin of Electrical Engineering and Informatics
Not a member yet
2885 research outputs found
Sort by
Optimal reactive power dispatch using modified-ant lion optimizer with flexible AC transmission systems devices
Get PDFThis study focuses on reactive power planning in the IEEE30-bus test system, specifically involving the integration of flexible AC transmission systems (FACTS) within the utility system. The primary objective is to minimize power loss and voltage deviation. To address this, a recently developed optimization algorithm called modified ant-lion optimizer (MALO) is applied to solve the optimal reactive power dispatch (ORPD) problem on the IEEE 30-bus system. A comparative analysis is conducted between the results obtained with and without FACTS devices. The findings reveal that the utilization of FACTS devices leads to significantly improved outcomes compared to scenarios without FACTS devices. Among the FACTS devices studied, the unified power flow controller (UPFC) demonstrates superior performance compared to the static synchronous compensator (STATCOM) and interline power flow controller (IPFC)
Advancement in self-powered implantable medical systems
Get PDFMany different elements of patient monitoring and treatment can be supported by implantable devices, which have proven to be extremely reliable and efficient in the medical profession. Medical professionals can use the data they collect to better diagnose and treat patients as a result. The devices’ power sources, on the other hand, are battery-based, which introduces a slew of issues. As part of this review, we explore the use of harvesters in implanted devices and evaluate various materials and procedures and look at how new and improved circuits can enable the harvesters to sustain medical devices
Deep learning approaches for analyzing and controlling rumor spread in social networks using graph neural networks
Get PDFThe pervasive influence of social networks on information dissemination necessitates robust strategies for understanding and mitigating the spread of rumors within these interconnected ecosystems. This research endeavors to address this imperative through the application of a graph neural network (GNN) model, designed to capture intricate relationships among users and content in social networks. The study integrates user-level attributes, content characteristics, and network structures to develop a comprehensive model capable of predicting the likelihood of rumor propagation. The proposed model is situated within a broader conceptual framework that incorporates sociological theories on information diffusion, user behavior, and network dynamics. The results of this research offer insights into the interpretability of the GNN model’s predictions and lay the groundwork for future investigations. The iterative refinement of the model, consideration of ethical implications, and comparison against traditional machine learning baselines emerge as crucial steps in advancing the understanding and application of deep learning methodologies for rumor control in social networks. By embracing the complexities of real-world scenarios and adhering to ethical standards, this research strives to contribute to the development of proactive tools for rumor management, fostering resilient and trustworthy online information ecosystems
Swift and efficient cinnamon plant disease classification using robust feature extraction and machine learning techniques
Get PDFThe extraction of features and textures holds a crucial significance in the realm of image processing and machine vision systems. Even though artificial intelligence (AI) techniques are superior in attaining the best results in image processing, several challenges remain open for further research in computation complexity, memory, and power requirements. In this context, robust preprocessing techniques are required to address such shortcomings and reduce the computational cost of predictive tasks. This paper employed two feature extraction levels to extract the best possible features from images of the cinnamon plant. Local directional positional pattern (LDPP) extracts global image features, while local triangular coded pattern (LTCP) extracts local features. It helps to provide detailed and more relevant information about the image texture. Once features are extracted, identifying and categorizing diverse textures within an image relies on recognizing their unique features. Typically, descriptors serve as the means for representing images in our work. Afterwards, we used ensemble learning to attain better classification results with the help of weak classifiers. Extracted features are provided to machine learning (ML) models like support vector machines (SVM), random forest (RF), and k-nearest neighbors (KNN) for better classification of the cinnamon category
Improving the performance of a U-shaped patch antenna using metamaterials for biomedical applications
Get PDFThis paper discusses the performance improvement of a patch antenna using metamaterials (MTM). The suggested antenna is a U-shaped patch antenna with a modified ground plane dedicated to biomedical applications. The size of the antenna is 40×20 mm2 with a FR4 substrate (εr=4.3, tanδ=0.02, H=1.6 mm) designed for operation at 2.4 GHz (ISM Band) and 6.23 GHz frequencies. The proposed MTM is 2×2 array positioned under the antenna at a distance of 2 mm. The integration of the MTM enhances clearly the antenna performance especially the return loss, voltage standing wave ratio (VSWR) and the gain. However, the reflection coefficient was enhanced from -10.71 dB to -36.63 dB at 2.45 GHz and from -13.88 dB to -36.54 dB at 6.23 GHz, the VSWR improved from 1.66 to 1.03 at 2.45 GHz and from 1.75 to 1.04 at 6.23 GHz. Additionally, the peak gain also was increased from 1.77 dB to 3.48 dB. The obtained results confirm the suitability of the suggested antenna for biomedical applications
Fusing Xception and ResNet50 features for robust grape leaf disease classification
Get PDFGrapes are one of the most widely cultivated fruits worldwide, and their economic and nutritional value makes them a significant crop in agriculture. However, grape plants are vulnerable to various diseases that can have detrimental effects on crop yield and quality. Accurate and timely identification of grape leaf diseases is crucial for efficient disease control and ensuring sustainable viticulture practices. In this study, we present a disease classification model specifically designed for grape leaves. The model incorporates bilinear pooling, utilizing the intermediate features extracted from two powerful convolutional neural network (CNN) models, Xception, and ResNet50. The outer product operation is applied to the extracted features, enabling the capture of intricate interactions and relationships between the features. The accurate classification of grape leaf diseases provided by our model offers significant benefits for grape farmers, vineyard owners, and agricultural researchers. It facilitates early disease detection, enabling proactive disease management strategies. Additionally, it assists in optimizing crop health, minimizing yield losses, and ensuring sustainable grape production
Exploring Bengali speech for gender classification: machine learning and deep learning approaches
Get PDFSpeech enables clear and powerful idea transmission. The human voice, rich in tone and emotion, holds unique beauty and significance in daily life. Vocal pitches vary by gender and are influenced by emotions and languages. While people naturally perceive these nuances, machines often struggle to capture these subtle distinctions. Machines may struggle to detect these nuances, but people effortlessly perceive them. This project aims to use various machine learning (ML) and deep learning (DL) techniques to reliably determine an individual’s gender from a corpus of Bengali conversations. Our dataset comprises 3185 Bengali speeches, with 1100 delivered by males, 1035 by women, and 1050 by those who identify as third gender. We employed six distinct feature extraction techniques to examine the audio data: roll-off, spectral centroid, chroma-stft, spectral bandwidth, zero crossing rate, and Mel-frequency cepstral coefficients (MFCC). Extreme gradient boosting (XGBoost), support vector machines (SVM), K-nearest neighbors (KNN), decision trees classifier (DTC), and random forest (RF) were employed as the five ML algorithms to comprehensively analyze the dataset. For a full study, we also included 1D convolutional neural networks (CNN) from the DL area. The 1D CNN performed extraordinarily well, exceeding the accuracy of all other algorithms with a stunning 99.37%
A hybrid extreme machine learning model for predicting heart disease
Get PDFHeart disease (HD), the leading cause of death for adults over 65, can affect anyone at any time. Additionally, modern lifestyles, poor diets, and other factors have led to an increased risk of HD among teenagers. One significant challenge is managing and analysing vast amounts of data, often surpassing terabytes, which is crucial for researching, diagnosing, and predicting cardiovascular diseases quickly. To enhance primary health care, especially in early and rapid diagnosis of heart attacks and to assist less experienced doctors in understanding clinical HD data, we propose a hybrid method called the "hybrid extreme machine learning model (HEMLM)". This technique combines the strengths of multi-layer perceptron (MLP), random layers, and logistic regression (LR). The model offers various feature patterns and multiple classification techniques. Compared to support vector machine (SVM), LR, and Naive Bayes (NB), the HEMLM algorithm demonstrates superior performance and efficiency. Testing results show identification accuracies of 94.91%, 94.77%, 92.42%, and 87.14% for data splitting ratios of 85:15, 80:20, 70:30, and 60:40, respectively
Optimized XGBRF-CatBoost model for accurate polycystic ovary syndrome prediction using ultrasound imaging
Get PDFPolycystic ovary syndrome (PCOS) is a multifactorial endocrine disorder characterized by hyperandrogenism, anovulation, oligomenorrhea, and ovarian microcysts, often resulting in infertility, obesity, and dermatological issues. This study proposes a hybrid machine learning (ML) framework for accurate PCOS prediction using ovarian ultrasound imaging and clinical parameters. A gradient regression-based multilayer perceptron neural network (GRMPNN) is employed for feature selection, followed by a stacked ensemble classifier combining extreme gradient boosted random forest (XGBRF) and CatBoost for final diagnosis. The dataset comprises 541 anonymized patient records from Ghosh Dastidar Institute for Fertility Research (GDIFR), incorporating 45 clinical, hormonal, and imaging features. Preprocessing includes normalization, noise reduction, and random oversampling to address class imbalance. Feature selection using univariate statistical testing and chi-square ranking identified 13 key attributes. The proposed XGBRF–CatBoost model achieved accuracy, precision, recall, and F1-score exceeding 98% across both benchmark datasets, outperforming principal component analysis (PCA) and neural fuzzy rough subset evaluating (NFRSE)-based models. This framework enhances diagnostic precision, reduces computational complexity, and supports scalable integration into clinical workflows. The findings underscore the potential of artificial intelligence (AI)-assisted tools in reproductive medicine and present a reproducible, interpretable approach for early PCOS detection
Adaptive AI-driven framework for digital mental health interventions in low-resource universities
Get PDFMental health problems affect nearly half of university students worldwide, with around 20% reporting depressive symptoms and over 40% showing signs of anxiety. This burden is particularly acute in low-resource universities, where limited infrastructure and minimal investment in mental health restrict access to effective care. To address this gap, this study applies a projective research approach, defined as the design of evidence-based solutions without immediate empirical implementation. A systematic review of 402 scientific articles was carried out across major databases, from which 15 met strict inclusion criteria. The analysis identified recurrent barriers such as unstable internet connectivity, devices with less than 2 GB RAM, and the absence of regulatory frameworks governing AI in education. Based on these findings, an adaptive intervention model was proposed, integrating artificial intelligence (AI), machine learning (ML), and deep learning (DL) to deliver personalized psychological support directly on local devices, without requiring permanent connectivity. The proposed system demonstrated potential to reduce anxiety and depression scores by 15–25% in controlled studies and achieved prediction accuracies above 80% for stress and loneliness detection. This framework provides a scalable foundation for universities in developing contexts, contributing to equity in access to digital mental health services