Izmir Institute of Technology

Permanent URI for this communityhttps://premium.gcris.co/handle/123456789/1

Browse

Now showing 1 - 4 of 4

High-Frequency Link Voltage Multiplexing for Multi-Level Inverters With Optimized Transformer Windings
(Institute of Electrical and Electronics Engineers Inc., 2025) Hataş, Hasan
The need for more than one voltage source in multilevel inverters (MLI) increases the system cost and circuit complexity. In this study, a voltage multiplexing method with a high frequency link (HFL) structure is proposed as a solution to the problem in question. Unlike the traditional HFL application, a half H -bridge is added to the input of the transformer and thus the obtained voltage is diversified. The output voltage is provided by a separate H-bridge circuit by preserving the isolation of the input and output voltages. The winding ratios of the transformer are determined by optimizing the total harmonic distortion (THD) of the output voltage with the NLC method. The proposed topology provides a low THD value of 1 1. 7 3% despite different voltage steps. In addition, simulations performed under various amplitude and frequency conditions using the SPWM technique have shown that the proposed method is effective and applicable. © 2025 Elsevier B.V., All rights reserved.
High-Frequency Link Voltage Multiplexing for Multi-Level Inverters With Optimized Transformer Windings
(Institute of Electrical and Electronics Engineers Inc., 2025) Hataş, Hasan
The need for more than one voltage source in multilevel inverters (MLI) increases the system cost and circuit complexity. In this study, a voltage multiplexing method with a high frequency link (HFL) structure is proposed as a solution to the problem in question. Unlike the traditional HFL application, a half H -bridge is added to the input of the transformer and thus the obtained voltage is diversified. The output voltage is provided by a separate H-bridge circuit by preserving the isolation of the input and output voltages. The winding ratios of the transformer are determined by optimizing the total harmonic distortion (THD) of the output voltage with the NLC method. The proposed topology provides a low THD value of 1 1. 7 3% despite different voltage steps. In addition, simulations performed under various amplitude and frequency conditions using the SPWM technique have shown that the proposed method is effective and applicable. © 2025 Elsevier B.V., All rights reserved.
A Lightweight Mobile Deep Learning Framework for Real-Time Plant Disease Detection in Smart Agriculture
(Institute of Electrical and Electronics Engineers Inc., 2025) Avcı, İsa; Koca, Murat; Khan, Yahya Zakrya
It is imperative to detect plant diseases early to enhance agricultural productivity and ensure food security. Conventional diagnostic techniques, which rely on the analysis of experts, are often laborious, expensive and less accessible, particularly in isolated regions. The present study proposes an automated plant disease detection system optimized using deep learning. The system utilizes Convolutional Neural Networks (CNNs). The proposed approach integrates advanced preprocessing techniques, including data augmentation, resizing, and normalization, to enhance model robustness and generalization. To facilitate deployment on mobile devices with limited resources, the MobileNetV2 architecture was optimized through quantization and conversion to TensorFlow Lite (TFLite). This approach resulted in a substantial reduction in computational complexity while maintaining an elevated level of classification accuracy. The mobile application, developed using Kotlin, facilitates the capture or upload of plant images and the execution of real-time disease detection directly on the device, thus obviating server communication. The experimental results demonstrate that the MobileNetV2 (Optimized) model achieved an accuracy of 99.48%, an F 1-score of 99%, and an AUC of 1.00, thus confirming its effectiveness for real-world agricultural applications. This study demonstrates the considerable potential of lightweight and efficient AI-driven solutions to transform the realm of plant disease detection, thereby rendering precision agriculture more accessible, particularly in resourceconstrained environments. © 2025 Elsevier B.V., All rights reserved.
A Lightweight Mobile Deep Learning Framework for Real-Time Plant Disease Detection in Smart Agriculture
(Institute of Electrical and Electronics Engineers Inc., 2025) Avcı, İsa; Koca, Murat; Khan, Yahya Zakrya
It is imperative to detect plant diseases early to enhance agricultural productivity and ensure food security. Conventional diagnostic techniques, which rely on the analysis of experts, are often laborious, expensive and less accessible, particularly in isolated regions. The present study proposes an automated plant disease detection system optimized using deep learning. The system utilizes Convolutional Neural Networks (CNNs). The proposed approach integrates advanced preprocessing techniques, including data augmentation, resizing, and normalization, to enhance model robustness and generalization. To facilitate deployment on mobile devices with limited resources, the MobileNetV2 architecture was optimized through quantization and conversion to TensorFlow Lite (TFLite). This approach resulted in a substantial reduction in computational complexity while maintaining an elevated level of classification accuracy. The mobile application, developed using Kotlin, facilitates the capture or upload of plant images and the execution of real-time disease detection directly on the device, thus obviating server communication. The experimental results demonstrate that the MobileNetV2 (Optimized) model achieved an accuracy of 99.48%, an F 1-score of 99%, and an AUC of 1.00, thus confirming its effectiveness for real-world agricultural applications. This study demonstrates the considerable potential of lightweight and efficient AI-driven solutions to transform the realm of plant disease detection, thereby rendering precision agriculture more accessible, particularly in resourceconstrained environments. © 2025 Elsevier B.V., All rights reserved.

Browse

Browsing Izmir Institute of Technology by Publisher "Institute of Electrical and Electronics Engineers Inc."