Yazar "Uluer, Ihsan" seçeneğine göre listele

Listeleniyor 1 - 11 / 11
  • Küçük Resim Yok
    Öğe
    Comparison of Free Space Measurement Using a Vector Network Analyzer and Low-Cost-Type THz-TDS Measurement Methods Between 75 and 325 GHz
    (Springer, 2017) Ozturk, Turgut; Morikawa, Osamu; Unal, Ilhami; Uluer, Ihsan
    Specifications of two measurement systems, free space measurement using a vector network analyzer and low-cost-type terahertz time-domain spectroscopy using a multimode laser diode, have been compared in the frequency region of millimeter/sub-THz waves. In the comparison, accuracy, cost, measurement time, calculation time, etc. were considered. Four samples (Rexolite, RO3003, Ultralam 3850HT-design, and L1000HF) were selected for the comparison of the specifications of the two methods. The acquired data was used to compute the complex permittivity of measured materials. The extracted results by free space measurement agreed well to the ones obtained by low-cost-type terahertz time-domain spectroscopy. This result proves free space measurement that can be assessed as a new method of material characterization in the sub-THz region successfully worked. Furthermore, free space measurement was proved to be suitable for a measurement in a narrow frequency range. On the other hand, low-cost-type terahertz time-domain spectroscopy has features not only low cost but also measurement capability in wide frequency range.
  • Küçük Resim Yok
    Öğe
    Development of extraction techniques for dielectric constant from free-space measured S-parameters between 50 and 170 GHz
    (Springer, 2017) Ozturk, Turgut; Elhawil, Amna; Uluer, Ihsan; Guneser, Muhammet Tahir
    This paper is a comprehensive study on Newton-Raphson technique used in millimeter wave frequencies for material characterization. Various algorithms are used for extracting the complex permittivity of a material from measured S-parameters. Efficiency of the methods depends on the initial guess and the accuracy of measured S-parameters for each thickness and frequency band. In this paper, we suggest the initial-value estimation method that helps to estimate a proper value for starting the algorithm. Moreover, an alternative extraction process is modelled that does not require new measured S-parameters or extracting process for each frequency band and thickness with a composed database. The estimation process is conducted partially as V (50-75 GHz), W (75-110 GHz), and D (110-170 GHz) frequency bands.
  • Küçük Resim Yok
    Öğe
    Development of Measurement and Extraction Technique of Complex Permittivity Using Transmission Parameter S 21 for Millimeter Wave Frequencies
    (Springer, 2017) Ozturk, Turgut; Hudlicka, Martin; Uluer, Ihsan
    This study provides an overview of measured S-parameters and its processing to extract the dielectric properties of materials such as Teflon, PMMA, and PVC which are preferred for materials characterization process. In addition, a correction model is presented for transmission parameter (S (21)) to obtain the dielectric constant with high accuracy. A non-destructive and non-contact free space measurement method has been used to measure S-parameters of thin samples in the low THz frequency range. S-parameters are measured in free space by vector network analyzer supported with two frequency extenders. Additionally, the parabolic mirrors are used to collimate the generated beam in wide frequency range. Furthermore, a standard filter process is performed to remove the undesired ripples in signal using singular spectrum analyzer before the implementation of extraction process. Newton-Raphson extraction technique is used to extract the material complex permittivity as a function of the frequency in Y-band (325-500 GHz).
  • Küçük Resim Yok
    Öğe
    Extracting the dielectric constant of materials using ABC-based ANNs and NRW algorithms
    (Taylor & Francis Ltd, 2016) Ozturk, Turgut; Elhawil, Amna; Dugenci, Muharrem; Unal, Ilhami; Uluer, Ihsan
    Five different Nicolson-Ross-Weir (NRW) extracting techniques are used to extract the dielectric constants of Teflon, Rexolite, Glass (borosilicate and soda-lime), Paper, and Ultralam 3850HT from S-parameters. The results of these extraction techniques are used to train the Artificial Neural Networks. In order to improve the accuracy of the results, the weights of ANNs are calculated using artificial bee colony estimation method. The results are compared with that obtained using NRW, Newton-Raphson, and genetic algorithm. The obtained results indicate that the proposed model gives good extracted parameters as compared with the previously published results.
  • Küçük Resim Yok
    Öğe
    An Investigation of the Effect of Embedded Gold Nanoparticles in Different Geometric Shapes on the Directivity of THz Photoconductive Antennas
    (Eos Assoc, 2023) Karasu, Yunus Emre; Uluer, Ihsan; Ozturk, Turgut
    This study aims to show how Terahertz (THz) Photoconductive Antennas (PCAs) affect the radiation directivity when gold nanoparticles of various geometric shapes are embedded in the gap region between the electrodes of THz PCAs. Three different PCAs, in the frequency range of 0.1 to 2 THz, conventional and after the addition of cylindrical, triangular, square, and hexagonal geometric gold nanoparticles to the antenna gap region between the electrodes, were built and simulated. The antenna directivity increased from 4.71 dBi to 4.92 dBi when square nanoparticles were added to the bowtie PCA, from 4.33 dBi to 4.45 dBi when triangular nanoparticles were added to the dipole PCA, and from 7.18 dBi to 7.52 dBi when square nanoparticles were added to the Vivaldi PCA.
  • Küçük Resim Yok
    Öğe
    Materials classification by partial least squares using S-parameters
    (Springer, 2016) Ozturk, Turgut; Uluer, Ihsan; Unal, Ilhami
    This paper presents a simple and usable algorithm, which is called Partial Least Squares, to classify the samples in different thicknesses and frequency range. This model employed to reconstruct the samples using S-parameters which are collected by Vector Network Analyzer in Free Space Measurement method. The relationship of between S-parameters is showed to classify the materials using proposed model. The Partial Least Squares algorithm has a good potential to classify the different samples with non-contactless and non-destructive measurement method. The classification process will be easy by using the proposed model, due to its practical and simple usage. In addition, the extraction techniques, which are Nicolson Ross Weir, Newton-Raphson, and Genetic Algorithm, are used in order to extract the dielectric constant of samples, in this study. The Newton-Raphson algorithm is selected to obtain the complex permittivity of samples since it gives best results than other extracting techniques and the results are used for classification.
  • Küçük Resim Yok
    Öğe
    Materials classification by partial least squares using S-parameters (vol 27, pg 12701, 2016)
    (Springer, 2019) Ozturk, Turgut; Uluer, Ihsan; Unal, Ilhami
    The original version of this article unfortunately published incorrectly in the definition of complex permittivity extraction given in the relation (5) of [1] due to its minus sign.
  • Küçük Resim Yok
    Öğe
    Optimization of Proportional-Integral Controllers of Grid-Connected Wind Energy Conversion System Using Grey Wolf Optimizer based on Artificial Neural Network for Power Quality Improvement
    (Lublin Univ Technology, Poland, 2022) Alremali, Fathi Abdulmajeed M.; Yaylaci, Ersagun kursat; Uluer, Ihsan
    This research presents a combination of artificial neural network (ANN) with the grey wolf optimizer (GWO) to improve the power quality of a grid-connected distributed power generation system (DPGS). To assess the effectiveness of the proposed algorithm, a grid-tied of small-scale wind energy conversion system (WECS) is chosen. The term power quality refers to voltage and frequency regulation, and limited harmonics. Power quality improvement is achieved through the cascaded control system's optimal tuning of three proportional-integral (PI) controllers of the grid-side inverter (GSI). However. because the DPGS model is computationally costly, the ANN model is utilized as an alternative model for DPGS. Furthermore, the ANN model is employed in conjunction with the GWO to boost the optimization precision and minimize the execution time of GWO. The considered power system was repetitively simulated to obtain the input-output datasets, which validate and train the ANN model. According to the ANN model's performance evaluation, the correlation coefficient (R) is close to one, while the mean squared error (MSE) is near zero. These findings demonstrate the ANN model's great accuracy in approximating the DPGS model. Using MATLAB/Simulink, the system's performance is evaluated using the optimum values obtained using GWO-ANN for various wind speed profiles. It showed the suggested power quality method's improved stability, convergence behavior, the effectiveness of the control mechanism, and the robustness of the proposed topology.
  • Küçük Resim Yok
    Öğe
    Reducing Speckle Noise from Ultrasound Images Using an Autoencoder Network
    (Ieee, 2020) Karaoglu, Onur; Bilge, Hasan Sakir; Uluer, Ihsan
    Image enhancement aims to obtain a clear image from a noisy image and it also uses for ultrasound images. In the experimental study, unlike classical image enhancement methods, deep learning method was used. Different levels of speckle noise added to the ultrasound images of the brachial plexus, which is known as the large nerve community under the armpit, were tried to be removed with the help of the convolutional denoising autoencoder network, which is one of the deep learning methods. The results obtained from the experimental study were compared with classical methods results and the proposed method was found to be more successful than classical methods.
  • Küçük Resim Yok
    Öğe
    Reduction of S-parameter errors using singular spectrum analysis
    (Amer Inst Physics, 2016) Ozturk, Turgut; Uluer, Ihsan; Unal, Ilhami
    A free space measurement method, which consists of two horn antennas, a network analyzer, two frequency extenders, and a sample holder, is used to measure transmission (S-21) coefficients in 75-110 GHz (W-Band) frequency range. Singular spectrum analysis method is presented to eliminate the error and noise of raw S-21 data after calibration and measurement processes. The proposed model can be applied easily to remove the repeated calibration process for each sample measurement. Hence, smooth, reliable, and accurate data are obtained to determine the dielectric properties of materials. In addition, the dielectric constant of materials (paper, polyvinylchloride-PVC, Ultralam (R) 3850HT, and glass) is calculated by thin sheet approximation and Newton-Raphson extracting techniques using a filtered S-21 transmission parameter. Published by AIP Publishing.
  • Küçük Resim Yok
    Öğe
    Removal of speckle noises from ultrasound images using five different deep learning networks
    (Elsevier - Division Reed Elsevier India Pvt Ltd, 2022) Karaoglu, Onur; Bilge, Hasan Sakir; Uluer, Ihsan
    Image enhancement methods are applied to medical images to reduce the noise that they contain. There are many academic studies in the literature using classical image enhancement methods. Ultrasound imaging is a medical imaging method that is used for the diagnosis of diseases. In this study, speckle noises with Rayleigh distribution at four different noise levels (sigma = 0.10, 0.25, 0.50, 0.75) are added to ultrasound images of the brachial plexus nerve region. Five different deep learning networks (Dilated Convolution Autoencoder Denoising Network/Di-Conv-AE-Net, Denoising U-Shaped Net/D-U-Net, BatchRenormalization U-Net/Br-U-Net, Generative Adversarial Denoising Network/DGan-Net, and CNN Residual Network/DeRNet) are used for reducing the speckle noises of the ultrasound images. The performances of the deep networks are compared with block-matching and 3D filtering (BM3D), which is one of the most preferred classical image enhancement algorithms; with classical filters including Bilateral, Frost, Kuan, Lee, Mean, and Median Filters; and with deep learning networks including Learning Pixel Distribution Prior with Wider Convolution for Image Denoising (WIN5-RB), Denoising Prior Driven Deep Neural Network for Image Restoration (DPDNN), and Fingerprint Image Denoising and Inpainting Using M-Net Based Convolutional Neural Networks (FPD-M-Net). Network performance is evaluated according to peak signal-to-noise ratio (PSNR), structural similarity index (SSIM), and runtime criteria and the proposed deep learning networks are shown to outperform the other networks. (C) 2021 Karabuk University. Publishing services by Elsevier B.V.