https://sjee.ftn.kg.ac.rs/index.php/sjee/issue/feedSJEE2026-11-17T17:49:57+00:00Prof. dr Alenka Milovanovićalenka.milovanovic@ftn.kg.ac.rsOpen Journal Systemshttps://sjee.ftn.kg.ac.rs/index.php/sjee/article/view/3507Time-domain Modelling of Low-power Transformer for Power Supply Applications2026-11-14T21:43:42+00:00Srdjan Divacsrdjan.divac@ftn.kg.ac.rsKrzysztof Chwastekkrzysztof.chwastek@pcz.plMariusz Najgebauermariusz.najgebauer@pcz.plBranko Koprivicabranko.koprivica@ftn.kg.ac.rs<p>The aim of this paper is to present a way of solving an electrical circuit of a single-phase transformer with a nonlinear hysteretic core in the time domain. The considered circuit consists of an AC power supply, a transformer with a ringshaped core and a resistor in the transformer primary and secondary circuit depicting the coil ohmic resistances. The secondary circuit also has a load impedance. Three working regimes of the transformer are considered – when it is loaded and when it is close to no-load and short-circuit working conditions. For the loaded regime, impedance has been varied from 10 Ω to 10 kΩ, with power factor from 0.01 to 0.99. In the no-load and short circuit regimes it is set to 10 MΩ and 0.01 Ω, respectively, with the power factor of 0.99. Calculations have been made at industrial frequency (50 Hz) using an iterative procedure based on Harmonic Interpolation Method (HIM) and Bertotti’s Statistical Theory of Losses (STL). The obtained results are compared with the results calculated using an equivalent transformer circuit and well-known RL model of the transformer core. The details regarding the calculation procedures, measurement setup and results, the comparison of the calculated results, as well as their appropriate analysis, are presented in this paper.</p>2026-11-14T20:00:54+00:00Copyright (c) 2026 SJEEhttps://sjee.ftn.kg.ac.rs/index.php/sjee/article/view/2868Dual-stage Enhanced Decision based Unsymmetric Trimmed Mean-Median Filter for High Density Salt and Pepper Noise Removal2026-11-14T21:39:08+00:00Aksha Perweenaksha.perween78@gmail.comAnshu Kumarianshu.kumari7424@gmail.comAbhishek Kumara.kumar1049@gmail.comSanjeev Kumarsanjeevsingh.ece@gmail.com<p>This paper presents a high-performance enhanced decision based unsymmetric trimmed mean-median filter (EDBUTMMF) to detect and remove high density salt and pepper noise from an image. Unlike conventional decisionbased filters that directly classify all extreme-valued pixels as noisy, the proposed method first verifies whether an extreme-valued pixel is genuinely corrupted before restoration. The process of denoising is implemented in two stages: noise detection followed by noise removal. The proposed filter initially detects noisy pixels iteratively through several phases and replaces the noisy pixels with the trimmed method. Unsymmetric trimmed filtering omits impulse noise values before performing the median calculation, ensuring that the replacement pixel is derived exclusively from the uncorrupted neighbourhood samples. This yields improved edge preservation and restoration accuracy, especially when dealing with high noise densities images. Dual-stage method separates noise detection and restoration to avoid unnecessary modification of uncorrupted pixels, especially at high density noise. Performance evaluation shows both quantitative parameters MSE, PSNR and SSIM achieves better results, while providing improved visual quality and lower computational time compared with several existing decisionbased filtering methods.</p>2026-11-14T21:20:58+00:00Copyright (c) 2026 SJEEhttps://sjee.ftn.kg.ac.rs/index.php/sjee/article/view/2900Illumination-Adaptive Edge Enhancement Using a Hybrid DSP–DL U-Net2026-11-14T21:39:10+00:00Milan Nikolicmisa_aum@hotmail.com<p>This paper presents a hybrid DSP–DL system for edge enhancement in images affected by severely degraded illumination, including under- and overexposed scenes. The proposed approach integrates deterministic gradientbased operators with an illumination-aware convolutional architecture to overcome the limitations of both classical edge extraction methods and learningbased detectors under non-uniform illumination conditions. A modified U-Net architecture is used to process a five-channel input tensor composed of histogramadjusted luminance, multiscale Sobel gradients (3×3 and 5×5), a map of global mean intensity, and a histogram-derived scene flag. Illumination adaptivity is introduced through a dynamic branching mechanism at the network input, to enable early modulation of feature extraction based on the luminance characteristics of the scene. A spatial attention module is introduced to support coherent edge reconstruction during decoding, while training of the network is guided by a dual loss formulation in which luminance consistency is balanced with gradient preservation through a learnable weighting parameter. An adaptive refinement stage further improves the edge continuity using luminance statistics and multiscale gradient cues. Quantitative evaluation is performed using five noreference (NR) edge and structure metrics, supported by statistical validation. Within the NR evaluation framework adopted here for illumination-degraded imagery, the proposed system demonstrates improved structural consistency relative to classical Sobel operators, and shows more stable behavior based on the NR metric when compared with representative pre-trained learning-based edge detectors under the same protocol and test conditions</p>2026-11-14T21:12:32+00:00Copyright (c) 2026 SJEEhttps://sjee.ftn.kg.ac.rs/index.php/sjee/article/view/2182A Multi-Objective Decision-Making Approach for Network Stability Enhancement by Reconfiguration and Installation of DG and Capacitor Units Using Crow Search Algorithm2026-11-14T21:39:11+00:00Ola Badrano.badran@ptuk.edu.psJafar Jalladj.jallad@ptuk.edu.ps<p>Network Reconfiguration (NR) is an efficient and essential methodology for minimizing power loss (Pl). However, NR is still critically challenging due to its complexity, nonlinearity, and large-scale optimization problem. This work introduces an innovative approach to enhance the Distribution Network (DN) stability and reliability. It addresses a simultaneous optimizing technique within different limits and constraints related to network power flow and voltage limits. It optimizes Distribution Reconfiguration (DR), Distributed Generation Sizing and Location (DG_SL), and shunt Capacitor Sizing and Location (CAP_SL). The main objectives are reducing Pl in the network, improving the Voltage Stability Index (VSI), and improving distribution efficiency. The Analytic-Hierarchy Process (AHP) and Crow Search Algorithm (CSA) are used as the foundation for a Multi-Objective Decision-Making Approach (MODMA). Simulations on IEEE 33 and 69 bus networks were run using MATLAB software to assess the viability of the suggested approach. This study obtains efficient output results and a positive impact on the distribution through the DR, DG_SL, and CAP_SL optimization. This appears through the system Pl reduction. For the IEEE 33-bus system, the reductions in Active Power Loss (Ploss) and Reactive Power Loss (Qloss) are 94.494% and 93.294%, respectively, whereas, for the IEEE 69-bus system, they are 97.097% and 94.412%, respectively. In contrast, the minimum bus voltage for the IEEE 33-bus system improved to 0.9926 p.u. and for the IEEE 69-bus system to 0.9912 p.u. For more validation, by comparing a part of the obtained results with other articles.</p>2026-11-14T21:32:13+00:00Copyright (c) 2026 SJEEhttps://sjee.ftn.kg.ac.rs/index.php/sjee/article/view/3447Hybrid Metaheuristic Method for Vegetation-Rich Outdoor 3D Point Cloud Image Registration2026-11-14T21:41:46+00:00Nebojša Čvorovićncvorovic3@gmail.comAna Gavrovskaanaga777@gmail.com<p>This paper presents a hybrid metaheuristic point cloud registration procedure combining Differential Evolution (DE) and Particle Swarm Optimization (PSO), referred to as DE-PSO. The method pairs the broad search capabilities of DE with the rapid convergence of PSO, avoiding the local minima traps that degrade standard Iterative Closest Point (ICP) implementations. Vegetation-rich outdoor environments present significant challenges due to dense foliage, irregular geometries, and severe point cloud noise. The proposed DE-PSO strategy addresses these complex scenes, achieving high registration accuracy and avoiding traditional ICP misalignments. Implemented using open-source C++ tools and tested on a custom dataset, the approach yields superior alignment with the lowest root mean square error and fastest execution time compared to standalone configurations. Overall, the method shows the feasibility of combining population-based metaheuristics for complex data, providing a practical tool for outdoor applications where conventional alignment struggles.</p>2026-11-14T21:03:21+00:00Copyright (c) 2026 SJEEhttps://sjee.ftn.kg.ac.rs/index.php/sjee/article/view/3492Numerically Enhanced Perturb-and-Observe Approach for Power Optimization in Wind Energy Systems2026-11-14T21:39:13+00:00Ivana Krstajićivanakrstajic0@gmail.comMartin Ćalasanmartinc@ucg.ac.meGojko Krunićgojko.krunic@fpm.ues.rs.ba<p>Modern power systems are facing increasing demands for decarbonization and a higher share of renewable energy sources, with wind energy systems playing an important role. However, the variable nature of wind complicates the stable and efficient operation of these systems, making the application of effective Maximum Power Point Tracking (MPPT) algorithms essential. Among the most commonly used methods is the Perturb and Observe (P&O) algorithm, characterized by its simplicity but limited by oscillations around the maximum power point and reduced efficiency under rapid wind speed variations. In this paper, a numerically enhanced P&O strategy with a variable perturbation step size based on different numerical approximations of the power–voltage (P-V) curve is proposed. The algorithm is applied to a wind energy conversion system with a Permanent Magnet Synchronous Generator (PMSG) and tested in the MATLAB/Simulink environment under various operating conditions. Simulation results demonstrate that the proposed approach provides faster dynamic response and higher generated energy values compared to the conventional P&O algorithm.</p>2026-11-14T20:07:56+00:00Copyright (c) 2026 SJEEhttps://sjee.ftn.kg.ac.rs/index.php/sjee/article/view/1617Effects of Rain Attenuation in 6G Communication2026-11-17T17:49:57+00:00Sharmini Enochsharminienoch@gmail.comIfiok Otungi.otung@bradford.ac.uk<p>In 6G communication the use of higher frequencies above 7 GHz to up to 10 THz is considered. Above the 10 GHz frequency range the atmospheric propagation effects such as rainfall can produce considerable attenuation in the received signal. The aim of this work is to demonstrate the effects of rain attenuation in high frequencies. The radar rain data from Chilbolton Facility for Atmospheric and Radio Research (CFARR) facility at Chilbolton, UK is utilized to analyze the magnitude of attenuation caused by rain attenuation. The results indicate that at an operating frequency of 56.5 GHz, the predicted specific rain attenuation leads to a total path attenuation of approximately 48 dB for 0.01% of the annual time. This exceedance probability corresponds to high-intensity rainfall events and represents a critical availability threshold for millimeter-wave communication systems. Due to the short wavelength at 56.5 GHz, electromagnetic wave propagation is highly susceptible to scattering and absorption by raindrops, resulting in pronounced signal degradation.</p>2026-11-14T21:38:30+00:00Copyright (c) 2026 SJEEhttps://sjee.ftn.kg.ac.rs/index.php/sjee/article/view/2554A Comprehensive Approach to Speech Emotion Recognition: Combining MFCC, Multifractal, Prosodic Features, and Volterra Filtering2026-11-14T21:39:14+00:00Fathima Kunhi Mohamedfathiphd2024@gmail.comAbdul Haleem Pezumkattil Pottammalabdulhaleem@kannuruniv.ac.in<p>This study proposes a new approach to speech emotion recognition by integrating traditional Mel-Frequency Cepstral Coefficients (MFCC) features with multifractal features derived from speech signals. Then a Sparse Adaptive Volterra Filter was used to improve these features. A Convolutional Neural Networks – Bidirectional Long Short- Term Memory (CNN-BiLSTM) was used to classify the emotions. Experiments were conducted on a self-built Malayalam emotional dataset of 3120 samples. Accuracy increased from 89.10% with the MFCC-only model to 93.43% when combined with multifractal features. This is an absolute improvement of 4.33%. Adding the Volterra filter raised the final accuracy to 95.51%, a total improvement of 6.41% over the MFCC-only baseline. Including pitch and formant features raised the accuracy further to 96.15%. For further validation, the method was also tested on the RAVDESS, a widely used benchmark dataset. Overall, the proposed framework gave high accuracy compared to baseline methods. These results show the potential of this methodology for practical emotion recognition applications, and points to the model’s robustness and effectiveness in real-world scenarios.</p>2026-11-14T21:26:45+00:00Copyright (c) 2026 SJEE