İclal Çetin Taş
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
İclal Çetin Taş has a total of 13 co-authors. Here are the top co-authors by number of shared publications:
- Sami Arica
Department of Electrical and Electronics Engineering, Cukurova University, Adana, Turkey
3 shared publications - Ahmet Anil Mungen
OSTIM Technical University,Software Engineering Department,Ankara,Turkey
1 shared publication - Ahmet Güngör Pakfiliz
Department of Electrical and Electronics Engineering, Turkish Aeronautical Association University Faculty of Engineering, Ankara, Türkiye
1 shared publication - Ali Can Ozdemir
Department of Mining Engineering, Çukurova University, Adana 01330, Türkiye
1 shared publication - Ali Musa Bozdoğan
Cukurova Universitesi
1 shared publication - Cagatay Cebeci
Department of Electrical and Electronics Engineering, Osmaniye Korkut Ata University, 80000 Osmaniye, Turkiye
1 shared publication
Latest Publications
- Open Access
MixTOPSIS: A TOPSIS Extension Incorporating Gower Distances for Selecting Unmanned Aerial Vehicles with Mixed-Scale Attributes
UAVs come in many different platforms such as quadcopters, hexacopters, hybrid models, etc., with diverse attribute types which can be characterised as heterogeneous data. In MCDM, heterogeneous datasets pose several challenges, as most methods are designed for homogeneous datasets. To address these challenges, this paper introduces the MixTOPSIS method by integrating the Gower dissimilarity metric into the TOPSIS framework, enabling the evaluation of alternatives with mixed data types, including numerical, ordinal, and nominal criteria. Therefore, to demonstrate how the method extends TOPSIS for mixed data types, the UAV ranking problem was chosen as the application. According to the simulation results, the top-ranked alternative consistently matched the preferred nominal criteria, while tare and maximum takeoff weight attributes were limiting elements for all options. Sensitivity analyses with gradual and random perturbations of criterion weights were conducted to further examine ranking stability. For the gradual perturbations, the dominant ranking pattern was maintained in 92.78% of the 1441 weight assignments. The analysis with random perturbations also showed substantial ranking stability, with the dominant pattern maintained in 98.02% of cases. The primary originality of MixTOPSIS lies in: (i) evaluating mixed-scale (nominal, ordinal, and numerical) criteria within the TOPSIS distance framework without requiring artificial numerical encoding; (ii) offering a flexible preferred-level mechanism for qualitative attributes; and (iii) introducing criterion-wise contribution metrics that enhance decision explainability. Overall, MixTOPSIS performed effectively, offering a robust, practically useful decision-making tool with interpretable results for complex real-world applications involving diverse data types.
Open AccessMixTOPSIS: A TOPSIS Extension Incorporating Gower Distances for Selecting Unmanned Aerial Vehicles with Mixed-Scale Attributes
UAVs come in many different platforms such as quadcopters, hexacopters, hybrid models, etc., with diverse attribute types which can be characterised as heterogeneous data. In MCDM, heterogeneous datasets pose several challenges, as most methods are designed for homogeneous datasets. To address these challenges, this paper introduces the MixTOPSIS method by integrating the Gower dissimilarity metric into the TOPSIS framework, enabling the evaluation of alternatives with mixed data types, including numerical, ordinal, and nominal criteria. Therefore, to demonstrate how the method extends TOPSIS for mixed data types, the UAV ranking problem was chosen as the application. According to the simulation results, the top-ranked alternative consistently matched the preferred nominal criteria, while tare and maximum takeoff weight attributes were limiting elements for all options. Sensitivity analyses with gradual and random perturbations of criterion weights were conducted to further examine ranking stability. For the gradual perturbations, the dominant ranking pattern was maintained in 92.78% of the 1441 weight assignments. The analysis with random perturbations also showed substantial ranking stability, with the dominant pattern maintained in 98.02% of cases. The primary originality of MixTOPSIS lies in: (i) evaluating mixed-scale (nominal, ordinal, and numerical) criteria within the TOPSIS distance framework without requiring artificial numerical encoding; (ii) offering a flexible preferred-level mechanism for qualitative attributes; and (iii) introducing criterion-wise contribution metrics that enhance decision explainability. Overall, MixTOPSIS performed effectively, offering a robust, practically useful decision-making tool with interpretable results for complex real-world applications involving diverse data types.
- Open Access
- 8 May 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (10) , 4676
A Novel Gene Expression Programming Algorithm for Forecasting Carbon Dioxide Emissions in G7 Countries
The increase in the carbon dioxide (CO2) emissions, nearly a quarter of those originating from the G7 countries, threatens not only the sustainability of the Earth but also the lives of future generations of humanity. Shedding light on future projections of the CO2 emissions is vital in achieving the target of carbon neutrality, and machine learning-based algorithms are frequently applied to forecast the CO2 emissions in the literature. However, the majority of these algorithms create model equations that are abstruse and irreproducible. In the current study, a novel gene expression programming (GEP) algorithm is proposed to produce genuine and easily understandable mathematical models for forecasting the CO2 emissions of the G7 countries. The proposed algorithm is comprehensively compared with both the simple GEP and the previous studies in terms of several error metrics and computational time. Consequently, the obtained results unveiled that the proposed algorithm surpassed the simple GEP by the improvements of 26% in nMAE, 24% in nRMSE, and 27% in MAPE, respectively. Notably, the proposed algorithm maintains essentially the same computational efficiency as the simple GEP (a 0.2% difference in duration) despite its richer function set. In addition to those, the estimated model equations belonging to the year of 2035 were meticulously presented to guide the researchers in the field for the sake of applicability and reproducibility.
- 8 May 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (10) , 4676
Open AccessA Novel Gene Expression Programming Algorithm for Forecasting Carbon Dioxide Emissions in G7 Countries
The increase in the carbon dioxide (CO2) emissions, nearly a quarter of those originating from the G7 countries, threatens not only the sustainability of the Earth but also the lives of future generations of humanity. Shedding light on future projections of the CO2 emissions is vital in achieving the target of carbon neutrality, and machine learning-based algorithms are frequently applied to forecast the CO2 emissions in the literature. However, the majority of these algorithms create model equations that are abstruse and irreproducible. In the current study, a novel gene expression programming (GEP) algorithm is proposed to produce genuine and easily understandable mathematical models for forecasting the CO2 emissions of the G7 countries. The proposed algorithm is comprehensively compared with both the simple GEP and the previous studies in terms of several error metrics and computational time. Consequently, the obtained results unveiled that the proposed algorithm surpassed the simple GEP by the improvements of 26% in nMAE, 24% in nRMSE, and 27% in MAPE, respectively. Notably, the proposed algorithm maintains essentially the same computational efficiency as the simple GEP (a 0.2% difference in duration) despite its richer function set. In addition to those, the estimated model equations belonging to the year of 2035 were meticulously presented to guide the researchers in the field for the sake of applicability and reproducibility.
- Open Access
- 6 April 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Wavelet-enhanced autoencoder based image denoising with CNN fusion
In biomedical imaging, noise is a fundamental problem negatively impacting diagnostic quality, and effective noise reduction methods are critical for preserving structural details. This study proposes a hybrid noise reduction framework integrated with a CNN-based fusion network, combining a convolutional autoencoder, a specially designed hybrid wavelet filtering approach, and multi-level adaptive wavelet transformation. The autoencoder component learns clean image matches from noisy inputs, while the wavelet-based arm ensures the preservation of high-frequency structural and textural information through multi-level decomposition. The CNN-based fusion module adaptively combines these complementary representations to produce the final enhanced image. Experimental results demonstrate that the proposed method provides consistent superiority across different noise types and datasets. In the fetal ultrasound dataset, the highest performance was achieved at all levels under Gaussian and speckle noise; for speckle noise low-level PSNR values of 48.12 dB and SSIM of 0.99 were obtained. For brain tumor MR dataset, the method was effective not only under Gaussian and speckle noise but also under Rician noise; the PSNR value was obtained as 22.90 dB at the low level, 27.11 dB at the medium level, and 22.17 dB at the high level, with corresponding SSIM values of 0.90, 0.59, and 0.47. Furthermore, in non-referenced evaluations performed on the wrist trauma clinical X-ray dataset, mean NIQE = 3.12 and SSEQ = 0.26 values were obtained. These results confirm that the proposed approach offers balanced and generalizable denoising performance without excessive smoothing or the creation of artificial artifacts on real clinical images. These findings demonstrate that combining wavelet-based detail preservation with deep learning-based reconstruction delivers robust and reliable noise reduction performance across diverse biomedical imaging modalities and noise types.
- 6 April 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Open AccessWavelet-enhanced autoencoder based image denoising with CNN fusion
In biomedical imaging, noise is a fundamental problem negatively impacting diagnostic quality, and effective noise reduction methods are critical for preserving structural details. This study proposes a hybrid noise reduction framework integrated with a CNN-based fusion network, combining a convolutional autoencoder, a specially designed hybrid wavelet filtering approach, and multi-level adaptive wavelet transformation. The autoencoder component learns clean image matches from noisy inputs, while the wavelet-based arm ensures the preservation of high-frequency structural and textural information through multi-level decomposition. The CNN-based fusion module adaptively combines these complementary representations to produce the final enhanced image. Experimental results demonstrate that the proposed method provides consistent superiority across different noise types and datasets. In the fetal ultrasound dataset, the highest performance was achieved at all levels under Gaussian and speckle noise; for speckle noise low-level PSNR values of 48.12 dB and SSIM of 0.99 were obtained. For brain tumor MR dataset, the method was effective not only under Gaussian and speckle noise but also under Rician noise; the PSNR value was obtained as 22.90 dB at the low level, 27.11 dB at the medium level, and 22.17 dB at the high level, with corresponding SSIM values of 0.90, 0.59, and 0.47. Furthermore, in non-referenced evaluations performed on the wrist trauma clinical X-ray dataset, mean NIQE = 3.12 and SSEQ = 0.26 values were obtained. These results confirm that the proposed approach offers balanced and generalizable denoising performance without excessive smoothing or the creation of artificial artifacts on real clinical images. These findings demonstrate that combining wavelet-based detail preservation with deep learning-based reconstruction delivers robust and reliable noise reduction performance across diverse biomedical imaging modalities and noise types.
- Open Access
- 17 March 2026
- Published by Anatolian Science - Bilgisayar Bilimleri Dergisi in Computer Science
- No. 2026
Performance Evaluation of Steganogan Using Hybrid Image Datasets and Statistical Quality Metrics
This study proposes a novel approach by employing SteganoGAN, a steganographic framework based on Generative Adversarial Networks (GANs). The impact of image categories with different semantic contents (art, building, and nature) on system performance was examined in detail using the SteganoGAN architecture. A hybrid dataset composed of BossBase and USC-SIPI images was utilized to conduct the experiments. The influence of image quality, categorized by visual content, on performance was systematically analyzed. In particular, changes in skewness and kurtosis values, as well as the widely used image quality metrics PSNR and SSIM, were investigated. Experimental results demonstrated that nature images provided superior performance, achieving an SSIM value of 99.59 and exhibiting the lowest skewness deviation from the original cover images.
- 17 March 2026
- Published by Anatolian Science - Bilgisayar Bilimleri Dergisi in Computer Science
- No. 2026
Open AccessPerformance Evaluation of Steganogan Using Hybrid Image Datasets and Statistical Quality Metrics
This study proposes a novel approach by employing SteganoGAN, a steganographic framework based on Generative Adversarial Networks (GANs). The impact of image categories with different semantic contents (art, building, and nature) on system performance was examined in detail using the SteganoGAN architecture. A hybrid dataset composed of BossBase and USC-SIPI images was utilized to conduct the experiments. The influence of image quality, categorized by visual content, on performance was systematically analyzed. In particular, changes in skewness and kurtosis values, as well as the widely used image quality metrics PSNR and SSIM, were investigated. Experimental results demonstrated that nature images provided superior performance, achieving an SSIM value of 99.59 and exhibiting the lowest skewness deviation from the original cover images.
- Open Access
- 26 June 2025
- Published by AVES YAYINCILIK A.Ş. in Electrica
- Vol. 25 (1) , 1-15
Cognitive Electronic Warfare Application to Emitter Identification Process in Airborne Radar Warning Receiver Systems
Cognitive electronic warfare (CEW) will soon replace conventional electronic warfare (EW) systems in the operational area. Although the CEW concept is utilized on various platforms, it has primarily emerged in military air platforms. Radar warning receiver (RWR) applications are essential in the CEW concept. Effective RWR systems accurately identify radar emitters in complex and dynamic environments, which is critical for situational awareness of air vehicle systems. systems provide information to countermeasure systems, increasing operational success by neutralizing enemy radar and defense systems. To sense the threats and apply appropriate countermeasures, such as jamming or dispensing chaff, requires a quick response and accurate RWR processes. RWR processes are summarized as receiving, signal detection, deinterleaving, labeling, and emitter identification. This study investigates the application of machine learning (ML) techniques to the emitter identification process. The research demonstrates significant improvements in emitter identification accuracy and distance determination by using ML algorithms. The findings suggest that CEW strategies incorporating adaptive learning models can enhance RWR system performance in identifying and categorizing emitters, even in dense Radio frequency (RF) environments. In the trials, it is observed that while the decision tree, ensemble bagged, and support vector machines achieved 100% success in the classification step, the Gaussian method was the most successful, with a mean absolute percent error of 1.6 in the distance calculation step using regression. These results suggest a promising direction for developing more intelligent and resilient EW systems. Cite this article as: İ. Çetin Taş and A. G. Pakfiliz, “Cognitive electronic warfare application to emitter identification process in airborne radar warning receiver systems,” Electrica, 2025, 25, 0182, doi: 10.5152/electrica.2025.24182.
- 26 June 2025
- Published by AVES YAYINCILIK A.Ş. in Electrica
- Vol. 25 (1) , 1-15
Open AccessCognitive Electronic Warfare Application to Emitter Identification Process in Airborne Radar Warning Receiver Systems
Cognitive electronic warfare (CEW) will soon replace conventional electronic warfare (EW) systems in the operational area. Although the CEW concept is utilized on various platforms, it has primarily emerged in military air platforms. Radar warning receiver (RWR) applications are essential in the CEW concept. Effective RWR systems accurately identify radar emitters in complex and dynamic environments, which is critical for situational awareness of air vehicle systems. systems provide information to countermeasure systems, increasing operational success by neutralizing enemy radar and defense systems. To sense the threats and apply appropriate countermeasures, such as jamming or dispensing chaff, requires a quick response and accurate RWR processes. RWR processes are summarized as receiving, signal detection, deinterleaving, labeling, and emitter identification. This study investigates the application of machine learning (ML) techniques to the emitter identification process. The research demonstrates significant improvements in emitter identification accuracy and distance determination by using ML algorithms. The findings suggest that CEW strategies incorporating adaptive learning models can enhance RWR system performance in identifying and categorizing emitters, even in dense Radio frequency (RF) environments. In the trials, it is observed that while the decision tree, ensemble bagged, and support vector machines achieved 100% success in the classification step, the Gaussian method was the most successful, with a mean absolute percent error of 1.6 in the distance calculation step using regression. These results suggest a promising direction for developing more intelligent and resilient EW systems. Cite this article as: İ. Çetin Taş and A. G. Pakfiliz, “Cognitive electronic warfare application to emitter identification process in airborne radar warning receiver systems,” Electrica, 2025, 25, 0182, doi: 10.5152/electrica.2025.24182.
- 19 June 2025
- Published by National Taiwan University in Biomedical Engineering: Applications, Basis and Communications
- Vol. 38 (02)
BRAIN TUMOR CLASSIFICATION OF MRI IMAGES VIA DEEP LEARNING APPROXIMATIONS: A REVIEW
In recent years, advancements in artificial intelligence and deep learning techniques have transformed various domains, including healthcare, by offering new ways to support clinical practices. These technologies are increasingly used in the diagnosis and treatment of diseases such as tumors, cancer, and neurological disorders. Among these, brain tumors are one of the most aggressive and damaging types of cancer. Due to the significant impact on the prognosis of a patient, early detection of brain tumors is crucial to improve treatment outcomes and monitoring treatment efficacy. Magnetic resonance imaging (MRI), a non-invasive imaging technique, plays a central role in diagnosing and evaluating brain tumors, providing detailed images critical for assessing tumor location, size, and progression. Deep learning models have achieved impressive results in medical image analysis, particularly in the classification of brain tumors using MRI. These models rely on large datasets of labeled medical images to identify complex patterns, textures, and features that distinguish different tumor types, such as gliomas or meningiomas. By training on these large datasets, deep learning models can learn subtle and high-level features that may be overlooked by human experts, thus improving diagnostic accuracy. This study aims to provide a comprehensive review of the current state of brain tumor classification using deep learning on MRI images. It seeks to explore the challenges researchers’ face, including data scarcity, model interpretability, and generalization issues across diverse datasets. Furthermore, it will highlight gaps in the existing literature and offer insights into future trends, such as hybrid models and multimodal imaging, that could enhance the accuracy and reliability of brain tumor diagnosis. The study aims to contribute valuable knowledge to medical image analysis by analyzing these aspects.
- 19 June 2025
- Published by National Taiwan University in Biomedical Engineering: Applications, Basis and Communications
- Vol. 38 (02)
BRAIN TUMOR CLASSIFICATION OF MRI IMAGES VIA DEEP LEARNING APPROXIMATIONS: A REVIEW
In recent years, advancements in artificial intelligence and deep learning techniques have transformed various domains, including healthcare, by offering new ways to support clinical practices. These technologies are increasingly used in the diagnosis and treatment of diseases such as tumors, cancer, and neurological disorders. Among these, brain tumors are one of the most aggressive and damaging types of cancer. Due to the significant impact on the prognosis of a patient, early detection of brain tumors is crucial to improve treatment outcomes and monitoring treatment efficacy. Magnetic resonance imaging (MRI), a non-invasive imaging technique, plays a central role in diagnosing and evaluating brain tumors, providing detailed images critical for assessing tumor location, size, and progression. Deep learning models have achieved impressive results in medical image analysis, particularly in the classification of brain tumors using MRI. These models rely on large datasets of labeled medical images to identify complex patterns, textures, and features that distinguish different tumor types, such as gliomas or meningiomas. By training on these large datasets, deep learning models can learn subtle and high-level features that may be overlooked by human experts, thus improving diagnostic accuracy. This study aims to provide a comprehensive review of the current state of brain tumor classification using deep learning on MRI images. It seeks to explore the challenges researchers’ face, including data scarcity, model interpretability, and generalization issues across diverse datasets. Furthermore, it will highlight gaps in the existing literature and offer insights into future trends, such as hybrid models and multimodal imaging, that could enhance the accuracy and reliability of brain tumor diagnosis. The study aims to contribute valuable knowledge to medical image analysis by analyzing these aspects.
- Open Access
- 30 April 2025
- Published by Bandirma Onyedi Eylul University in Mühendislik Bilimleri ve Araştırmaları Dergisi
- Vol. 7 (1) , 41-50
ELA (Error Level Analysis) kullanarak CNN Tabanlı Görüntü Sahteciliği Tespiti
Günümüzde görüntü sahteciliği, dijital ortamda bilgi güvenliğini tehdit eden önemli bir sorun haline gelmiştir. Dijital medyanın hızla yayılmasıyla birlikte, görüntü manipülasyonları artış göstermiş ve güvenilir bilgi paylaşımını zorlaştırmıştır. Bu bağlamda, görüntü sahteciliği, dijital çağın en önemli bilgi güvenliği sorunlarından biri olarak öne çıkmaktadır. Bu çalışmada, Hata Seviyesi Analizi (ELA) ve Derin Sinir Ağları (CNN) kullanılarak görüntü sahteciliğinin tespit edilmesi amaçlanmıştır. İlk olarak ELA'nın temel prensipleri ve uygulama yöntemleri ele alınmış, ardından CNN'in mimarisi ve eğitim süreçleri detaylandırılmıştır. Casia, Columbia ve hibrit veri setleri üzerinde gerçekleştirilen deneyler sonucunda, önerilen yöntemin yüksek doğruluk oranlarına ulaştığı gözlemlenmiştir. Elde edilen bulgular, ELA ve CNN kombinasyonunun görüntü sahteciliği tespitinde etkili bir araç olduğunu ortaya koymaktadır. Deneyler sonunda CNN vee la beraber kullanıldığından en yüksek F1-skor değeri 0.94 bulunmuştur.
- 30 April 2025
- Published by Bandirma Onyedi Eylul University in Mühendislik Bilimleri ve Araştırmaları Dergisi
- Vol. 7 (1) , 41-50
Open AccessELA (Error Level Analysis) kullanarak CNN Tabanlı Görüntü Sahteciliği Tespiti
Günümüzde görüntü sahteciliği, dijital ortamda bilgi güvenliğini tehdit eden önemli bir sorun haline gelmiştir. Dijital medyanın hızla yayılmasıyla birlikte, görüntü manipülasyonları artış göstermiş ve güvenilir bilgi paylaşımını zorlaştırmıştır. Bu bağlamda, görüntü sahteciliği, dijital çağın en önemli bilgi güvenliği sorunlarından biri olarak öne çıkmaktadır. Bu çalışmada, Hata Seviyesi Analizi (ELA) ve Derin Sinir Ağları (CNN) kullanılarak görüntü sahteciliğinin tespit edilmesi amaçlanmıştır. İlk olarak ELA'nın temel prensipleri ve uygulama yöntemleri ele alınmış, ardından CNN'in mimarisi ve eğitim süreçleri detaylandırılmıştır. Casia, Columbia ve hibrit veri setleri üzerinde gerçekleştirilen deneyler sonucunda, önerilen yöntemin yüksek doğruluk oranlarına ulaştığı gözlemlenmiştir. Elde edilen bulgular, ELA ve CNN kombinasyonunun görüntü sahteciliği tespitinde etkili bir araç olduğunu ortaya koymaktadır. Deneyler sonunda CNN vee la beraber kullanıldığından en yüksek F1-skor değeri 0.94 bulunmuştur.
- Open Access
- 28 March 2025
- Published by Istanbul Gedik University in International Journal of New Findings in Engineering, Science and Technology
- Vol. 3 (1) , 32-43
Estimation of Battery Remaining Life-time with Deep Learning Methods
The swift proliferation of renewable energy sources and electric grids causes discrepancies between energy supply and demand. This scenario causes variations in voltage and frequency levels due to discrepancies between energy generation and consumption, jeopardizing the stability of energy networks. The intrinsically fluctuating and unpredictable characteristics of renewable energy sources, such as the sun and wind, intensify these oscillations. In contrast to conventional have to have energy-producing systems, renewable energy systems have energy-producing systems and a restricted ability to adapt immediately to demand. In this environment, energy storage devices arise as a vital solution for the effective management of renewable energy generation and for sustaining grid stability. Research on Remaining Useful Life (RUL) and State of Charge (SoC) of batteries is essential for battery reliability, user satisfaction, and environmental sustainability. These studies provide benefits in energy efficiency, increased mobility, diminished battery replacement requirements, and superior waste management. Estimating battery longevity facilitates the efficient management of battery-operated equipment and the strategic planning of energy requirements. Deep learning techniques have made substantial progress in estimating battery capacity and longevity. Long-lasting batteries with substantial energy storage capacity, favored in industrial applications, are more efficiently assessed utilizing deep learning methodologies. This study analyzes the outcomes derived from the application of the Scaled Conjugate Gradient (SCG) technique for estimating battery capacity. It seeks to enhance the efficient management of battery systems and devise strategies that promote the sustainability of energy storage technology. This study's performance measures, comprising 1.098% MAPE, 0.9823 R², 0.0019 MSE, and 0.0302 MAE, enhance the effective management of energy storage systems, the optimal use of energy resources, and strategic planning to fulfill energy demands. This study's performance measures, 1.098% MAPE, 0.9823 R2, 0.0019 MSE and 0.0302 MAE obtained in this study on battery estimation, it supports the efficient management of energy storage systems, effective use of energy resources and strategic planning for energy demands.
- 28 March 2025
- Published by Istanbul Gedik University in International Journal of New Findings in Engineering, Science and Technology
- Vol. 3 (1) , 32-43
Open AccessEstimation of Battery Remaining Life-time with Deep Learning Methods
The swift proliferation of renewable energy sources and electric grids causes discrepancies between energy supply and demand. This scenario causes variations in voltage and frequency levels due to discrepancies between energy generation and consumption, jeopardizing the stability of energy networks. The intrinsically fluctuating and unpredictable characteristics of renewable energy sources, such as the sun and wind, intensify these oscillations. In contrast to conventional have to have energy-producing systems, renewable energy systems have energy-producing systems and a restricted ability to adapt immediately to demand. In this environment, energy storage devices arise as a vital solution for the effective management of renewable energy generation and for sustaining grid stability. Research on Remaining Useful Life (RUL) and State of Charge (SoC) of batteries is essential for battery reliability, user satisfaction, and environmental sustainability. These studies provide benefits in energy efficiency, increased mobility, diminished battery replacement requirements, and superior waste management. Estimating battery longevity facilitates the efficient management of battery-operated equipment and the strategic planning of energy requirements. Deep learning techniques have made substantial progress in estimating battery capacity and longevity. Long-lasting batteries with substantial energy storage capacity, favored in industrial applications, are more efficiently assessed utilizing deep learning methodologies. This study analyzes the outcomes derived from the application of the Scaled Conjugate Gradient (SCG) technique for estimating battery capacity. It seeks to enhance the efficient management of battery systems and devise strategies that promote the sustainability of energy storage technology. This study's performance measures, comprising 1.098% MAPE, 0.9823 R², 0.0019 MSE, and 0.0302 MAE, enhance the effective management of energy storage systems, the optimal use of energy resources, and strategic planning to fulfill energy demands. This study's performance measures, 1.098% MAPE, 0.9823 R2, 0.0019 MSE and 0.0302 MAE obtained in this study on battery estimation, it supports the efficient management of energy storage systems, effective use of energy resources and strategic planning for energy demands.
- Open Access
- 31 January 2025
- Published by El-Cezeri: Journal of Science and Engineering in El-Cezeri Fen ve Mühendislik Dergisi
- Vol. 12 (1) , 74-85
Classification of Dementia Levels by Using Different Convolutional Neural Network Architectures
Dementia or Alzheimer is a disease that causes symptoms such as forgetfulness and loss of physical ability, which will add to the individual's life in later stages, along with morphological changes in the brain. Unfortunately, a definitive treatment for these diseases has not yet been found. However, it is aimed at slowing down the progression of the disease to ensure that the patient is less affected by these adverse conditions and to protect living standards with early diagnosis of the disease. In addition, a complete diagnosis of the disease requires a series of tests and a tiring diagnostic phase to be evaluated by an experienced specialist. High-resolution magnetic resonance imaging is used to make this determination. This study tries to determine the stage of the disease or whether the individual is healthy by using MR.MR images of individuals in 4 stages of the disease, one of which is a healthy individual, were described as a classification problem and tried to be solved using VGG, Resnet, and Mobilenet architectures. Over 95% success has been achieved by supporting the proposed architecture with feature analysis and classical architectures.
- 31 January 2025
- Published by El-Cezeri: Journal of Science and Engineering in El-Cezeri Fen ve Mühendislik Dergisi
- Vol. 12 (1) , 74-85
Open AccessClassification of Dementia Levels by Using Different Convolutional Neural Network Architectures
Dementia or Alzheimer is a disease that causes symptoms such as forgetfulness and loss of physical ability, which will add to the individual's life in later stages, along with morphological changes in the brain. Unfortunately, a definitive treatment for these diseases has not yet been found. However, it is aimed at slowing down the progression of the disease to ensure that the patient is less affected by these adverse conditions and to protect living standards with early diagnosis of the disease. In addition, a complete diagnosis of the disease requires a series of tests and a tiring diagnostic phase to be evaluated by an experienced specialist. High-resolution magnetic resonance imaging is used to make this determination. This study tries to determine the stage of the disease or whether the individual is healthy by using MR.MR images of individuals in 4 stages of the disease, one of which is a healthy individual, were described as a classification problem and tried to be solved using VGG, Resnet, and Mobilenet architectures. Over 95% success has been achieved by supporting the proposed architecture with feature analysis and classical architectures.
- 20 January 2025
- Published by Turkish Journal of Engineering in Turkish Journal of Engineering
- Vol. 9 (1) , 1-11
Application of Convolutional Neural Networks for Watermelon Detection in UAV Aerial Images: A Case Study
Although human dependence on agriculture decreases with developing technology, it continues. As many resources are increasingly restricted due to various climatic reasons, the importance of studies in this field increases. Applications using deep learning models are frequently encountered in the agricultural field. In particular, there are applications where deep learning models are used as a tool for optimum planting, land use, yield improvement, production/disease/pest control, and other activities.In this study, watermelons in an aerial view of a watermelon field were detected by utilizing the Alexnet deep learning architecture. To obtain yield, watermelons in watermelon fields should be specified and then counted. Aerial images are used for this application. The field image was divided into 50% overlapping sub-images, and each was classified as watermelon, leaf, and soil. Consequently, watermelon regions on the field image were specified. After training the Alexnet and Vgg19 network structure with the dataset, watermelons were to be identified by segmenting the images. It was observed that the Vgg19 network achieved 97.78% accuracy. The results of the experimental applications show that the Vgg19 can be applied for watermelon fruit and yield detection applications.
- 20 January 2025
- Published by Turkish Journal of Engineering in Turkish Journal of Engineering
- Vol. 9 (1) , 1-11
Application of Convolutional Neural Networks for Watermelon Detection in UAV Aerial Images: A Case Study
Although human dependence on agriculture decreases with developing technology, it continues. As many resources are increasingly restricted due to various climatic reasons, the importance of studies in this field increases. Applications using deep learning models are frequently encountered in the agricultural field. In particular, there are applications where deep learning models are used as a tool for optimum planting, land use, yield improvement, production/disease/pest control, and other activities.In this study, watermelons in an aerial view of a watermelon field were detected by utilizing the Alexnet deep learning architecture. To obtain yield, watermelons in watermelon fields should be specified and then counted. Aerial images are used for this application. The field image was divided into 50% overlapping sub-images, and each was classified as watermelon, leaf, and soil. Consequently, watermelon regions on the field image were specified. After training the Alexnet and Vgg19 network structure with the dataset, watermelons were to be identified by segmenting the images. It was observed that the Vgg19 network achieved 97.78% accuracy. The results of the experimental applications show that the Vgg19 can be applied for watermelon fruit and yield detection applications.