Browsing by Author "Dr. K.M.A. Salam"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Open Access
    Automatic Fault Detection in Three-phase Transmission Lines
    (North South University, 2023) Md. Rakibul islam; Md. Sadaf islam; Imran islam; Dr. K.M.A. Salam; 1813378643; 1811681643; 1811457043
    This report presents the prototype design and implementation of ‘Automatic Fault Detection in Three-phase Transmission Lines’ using Arduino as microcontroller. The goal of the proposed Transmission Line Fault Detection System is to deliver an automated fault detection solution that can quickly and accurately identify and locate faults on transmission lines. The primary objective of this is to establish a rapid and precise identification and localization of faults within transmission lines, ultimately ensuring a seamless and uninterrupted flow of electricity. A prototype of this scheme is developing and testing on a simulated transmission line setting to confirm the system’s performance. The project provides an automatic tripping mechanism for the three-phase supply system to prevent defective damage. Various fault scenarios were artificially introduced to evaluate the system's ability to accurately detect and locate faults. This process allows for a comprehensive assessment of the system’s efficiency and reliability. Focus of this project was to enhance the reliability of three-phase transmission lines by employing voltage imbalance and phase imbalance factors for fault detection. Detailed analysis and calibration of these factors enable the system to swiftly identify faults and initiate timely corrective actions ensuring the uninterrupted flow of electricity within power networks. This proactive approach minimizes disruptions and potential damage, contributing to improved grid stability and efficiency.
  • Thumbnail Image
    Item
    Open Access
    Detection of Alzheimer’s Disease using Deep Learning
    (North South University, 2023) Md. Saron Ahmed; Sharmin Akter Mukti; Salman Sad Shakil; Chayan Banik; Dr. K.M.A. Salam; 1821641642; 1831029642; 1711064642; 1521170642
    Over the past many years, Machine Learning has been at its peak. It has many uses, such as predictive online browsing, email and text classification, detecting objects, and recognizing faces. Deep learning has grown in prominence during the past several years relative to all other machine learning applications. It helps researchers solve problems in the field of biomedicine, such as finding cancer, Alzheimer's, and malaria and figuring out what kind of blood cell something is. Deep learning is a subset of machine learning techniques used to pull out features for classification, image processing, etc. Using Magnetic Resonance Imaging (MRI) data, we classified Alzheimer patients from healthy patients using a Convolutional Neural Network (CNN). The OASIS-1 dataset has 416 people with Alzheimer's disease ranging from mild to moderate severity. Classifying this medical data is essential for making a prediction model or system that can tell if an infection is present in different people or what stage it is in. Alzheimer's disease has always been hard to put into a category, and figuring out what makes it different is the hardest part of this process. We have distinguished Alzheimer's patients from healthy participants using MRI data and various CNN architectures, including InceptionV3, Resnet50, MobileNetV2, VGG16, and VGG19, by calculating model accuracy, confusion matrix, and ROC curve. The most accurate models are the basic CNN and the InceptionV3, with an accuracy of up to 90.62 percent. This research demonstrates the performance of various CNN architectures on our MRI data of Alzheimer's patients and healthy participants in terms of classification. It enables us to identify the most effective models for detecting Alzheimer's disease.