This paper addresses the smart management and control of an independent hybrid system based onrenewable energies. The suggested system comprises a photovoltaic system (PVS), a wind energyconversion system (WECS), a battery storage system (BSS), and electronic power devices that arecontrolled to enhance the efficiency of the generated energy. Regarding the load side, the systemcomprises AC loads, DC loads, and a water pump. An Adaptive Neuro-Fuzzy Inference System (ANFIS)-based MPPT technique is suggested to enhance the efficiency of the PVS and WECS. This technologyprovided good performance compared with the Perturb and Observe (P&O) algorithm and MPPT-basedfuzzy logic controller (FLC). The use of the ANFIS-PI proposed to control the bidirectional converteraccomplished voltage stabilization for the DC bus. This work also came with a fuzzy logic-basedalgorithm to manage the load side that depends on battery charge ratio, solar radiation, and windspeed. According to results obtained in the MATLAB/Simulink environment, the proposed technologieswere found to have performed well. The goal we were also pursuing was achieved through the fulluse of the energy generated by the proposed algorithm. The proposed study holds great potential forremote regions.Index terms: Renewable energy, Hybrid system, MPPT, ANFIS controller, Management

While Model Predictive Control (MPC) has been widely studied in power electronics, its real-time imple-mentation on DC-DC boost converters—particularly under variable loading conditions—remains limited. Thispaper proposes a new real-time implementation of the Model Predictive Control (MPC) law for a DC-DC boostconverter connected to variable loads. This implementation ensures precise current regulation through accurateduty cycle control updates, enabling the inverter’s frequency switching to be activated or deactivated as needed.This is achieved by proposing a predictive model of the current occurring in the first channel of the convertermodel, where a fitness function—comprising reference tracking and control effort—is minimized. Compared tothe proportional-integral (PI) controller, the MPC law proves more efficient, particularly in preventing oscilla-tions in both transient and steady-state output current responses. This advantage is validated through experi-mental tests for either a current inductance load or a resistive load. Since this type of real-time implementationhas not been previously applied on this converter, it constitutes the main contribution of this paper.
Keywords: PI controller, DC-DC boost converters, Model predictive control (MPC), Experimental validation

This paper provides an in-depth analysis and performance evaluation of four Solar Radiance (SR)prediction models. The prediction is ensured for a period ranging from a few hours to several days ofthe year. These models are derived from four machine learning methods, namely the Feed-forwardBack Propagation (FFBP) method, Convolutional Feed-forward Back Propagation (CFBP) method,Support Vector Regression (SVR), and the hybrid deep learning (DL) method, which combinesConvolutional Neural Networks and Long Short-Term Memory networks. This combination results inthe CNN-LSTM model. Additionally, statistical indicators use Mean Squared Error (MSE), Root MeanSquared Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), andNormalized Root Mean Squared Error (nRMSE). Each indicator compares the predicted output by eachmodel above and the actual output, pre-recorded in the experimental trial. The experimental resultsconsistently show the power of the CNN-LSTM model compared to the remaining models in terms ofaccuracy and reliability. This is due to its lower error rate and higher detection coefficient (R2 = 0.99925).Keywords: Artificial neural networks, Convolutional neural network, Convolutional feed-forward backpropagation, Deep learning, Feed-forward back propagation, Long short-term memory, Solar radianceforecasting

Integration of nonlinear loads in modern power systems has led to many issues arising mainly dueto the generation of harmonic currents and the presence of reactive power, both having adverseeffects on power quality and grid stability. Harmonic currents cause increased losses, overheatingof equipment, and voltage distortions, while reactive power imbalances result in inefficiencies inpower delivery and compromised system performance. To overcome these problems, a Shunt ActivePower FIlter design and an optimal control strategy for harmonic mitigation and reactive powercompensation are proposed in this paper. The design incorporates an optimized anti-windup PIcontroller for DC-link voltage regulation and an optimized output filter to enhance the quality of theinjected current. This design is formulated as an optimization problem and solved using the GoldenJackal Optimizer. MATLAB/Simulink simulations validate the proposed method under differentoperating conditions, covering dynamic change of loads and unbalanced grid conditions. The resultshows a remarkable reduction in Total Harmonic Distortion (THD) of grid current, and reactive powercompensation meanwhile maintaining the stability of the grid.Keywords: Golden Jackal optimization, Shunt active power filter (SAPF), Optimal control, Power quality,Current harmonics compensation

The work presented in this paper focuses on formulating the development of time-dependent electromagneticfield laws through the application of Gauss’s divergence theorem. The first part of the discussion looks at thebasic ideas of electromagnetism. It focuses on how classical formulations of the laws of electromagnetism can beadapted to account for non-stationary conditions, especially regarding magnetic fluids that don’t conduct elec-tricity. It is suggested that employing Gauss’s divergence theorem could help improve the computational analysisof these generalized equations, which would make them more useful in magnetic fluid dynamics. The paperexamines the intricate interactions between non-conductive particles and conductive fluids under magneticfields. By putting these interactions into a single theoretical framework, this work aims to help us understandnon-stationary electromagnetic phenomena and how they affect many different scientific and engineering fields.The concluding section of the study examines the prospective practical applications of these extended equations.They could enable the development of more advanced electromagnetic devices and systems. Creating a strong setof analytical tools that can find new scientific paths and useful applications is the main goal of the study,particularly in the areas of electromagnetic induction and fluid dynamics. This research offers potential forsubstantial progress in both theoretical comprehension and technological advancement, The proposed method isapplicable to real-world systems such as ferrofluid-based cooling, magnetic dampers, plasma generators, andsmart electromagnetic devices. These applications demonstrate the practical benefits of coupling field behaviorwith boundary dynamics using Gauss’s theorem.
Keywords: Gauss theorem, Non-conductive;Magnetic, Non-stationary, Fluids, Induction

Hate speech has been disseminated more frequently on social media sites like Facebook in recent years. OnFacebook, hate speech can proliferate through text, image, or video. We suggested a deep learning approach toidentify offensive memes posted on Facebook in case of Amharic language’. The research process commenced bymanually gathering memes posted by Facebook users. Next came textual data extraction, annotation, pre-processing, splitting, feature extraction, model development and assessment Amharic OCRs were employed toextract textual data. Character normalization, stop word removal, and unnecessary character removal make upthe text-preprocessing step. Using Stratified KFold the textual dataset is split into the train set (80 %), thevalidation set (10 %) and the test set (10 %). Vectors are created from the preprocessed texts using the Bog ofwords (BOW), TFIDF and word embeddings. Following that, the vectors are fed into Machine learning algo-rithms: NB, DT, RF, KNN, LSVM and LR, and deep learning models that are based on Dense, BiGRU, and BiLSTMalgorithms. The model with the optimal parameters is chosen after numerous experiments. With an accuracy rateof 94 %, the BiLSTM + Dense model, the suggested technique identified nasty meme posts on Facebook written inAmharic.
Keywords: Deep learning, BILSTM, BIGRU, Amharic language hate speech

Object tracking is a technique for finding moving objects of interest and estimating their trajectoryor path with regard to time in a series of images. It involves object representation, detection,and tracking. It becomes an important field of study due to the need in video surveillance, trafficmonitoring, live sport video analysis and many other applications. In this paper, both static camera-based and dynamic camera-based object tracking techniques have been developed. The static camera-based object tracking was developed with NI LabVIEW, and Shape adaptive mean-shift algorithmhas been used for tracking. In case of dynamic camera-based object tracking, an optimal Fuzzy-PIDcontroller has been designed to adjust the position of the pan/tilt mechanism so as to trace the object’strajectory. Genetic algorithm (GA) was used to find the optimal values of the operating ranges (scalingfactors) of the membership functions. The performance of the system has been tested by differenttrajectories like step, sinusoidal, circular and elliptical at different frequencies 1, 50 and 100 rad/sec.The system has best performance at low frequencies and when the frequency or speed of the objectincreases, the system performance decreases which complies for real systems. The simulation resultsdemonstrate that GA tuned Fuzzy-PID controller has given us the best results in terms of reducedsteady-state error, faster rise time and settling time, and object position stabilization than PID,Fuzzy and Fuzzy-PID controllers, which shows that optimal Fuzzy-PID controller designed is moreappropriate and efficient.Keywords: Object tracking, LabVIEW, Fuzzy-PID, Pan/Tilt system, Genetic algorithm

In engineering applications, the bearing faults diagnosis is essential for maintaining reliability andextending the lifespan of rotating machinery, thereby preventing unexpected industrial productiondowntime. Prompt fault diagnosis using vibration signals is vital to ensure seamless operation ofindustrial system avert catastrophic breakdowns, reduce maintenance costs, and ensure continuousproductivity. As industries evolve and machines operate under diverse conditions, traditional faultdetection methods often fall short. In spite of significant research in recent years, there remains apressing need for improve existing methods of fault diagnosis. To fill this research gap, this researchwork aims to propose an efficient and robust system for diagnosing bearing faults, using deep andShallow features. Through the evaluated experiments, our proposed model Multi-Block Histogramsof Local Phase Quantization (MBH-LPQ) showed excellent performance in classification accuracy, andthe audio-trained VGGish model showed the best performance in all tasks. Contributions of this workinclude: Combine the proposed Shallow descriptor, derived from a novel hand-crafted discriminativefeatures MBH-LPQ, with deep features obtained from VGGish pre-trained of Convolutional NeuralNetwork (CNN) using audio spectrograms, by merging at the score level using Weighted Sum (WS).This approach is designed to take advantage of the complementary strengths of both feature models,thus enhancing overall bearing fault diagnostic performance. Furthermore, experiments conductedto verify the approach’s performance is assessed based on fault classification accuracy demonstrateda significant accuracy rate on two different noisy datasets, with an accuracy rate of 98.95% and 100%being reached on the CWRU and PU datasets benchmark, respectively.Keywords: Bearing fault diagnosis, Vibration signals, Transfer learning, Shallow descriptor, Deep features,MBH-LPQ, VGGish, CNN

Accurate regulation of the liquid level in a quadruple tank system (QTS) is not easy and imposes higherrequirements on control strategies, so the design of controllers in these systems is challenging dueto the difficulty of dynamic analysis of its nonlinear characteristics and parametric uncertainties.To overcome these problems in liquid level regulation and increase the robustness to the pumpcoefficients, this article proposes and investigates the use of an optimal hybrid fractional-ordertype-2 fuzzy-PID (OH-FO-T2F-PID) regulator using a combination of two bio-inspired evolutionaryoptimizers, namely augmented grey wolf optimizer and cuckoo search optimizer, which gives rise tothe new hybrid A-GWOCS algorithm. This control mechanism was chosen to facilitate the convergenceof the water liquids in the two tanks as quickly as possible to the corresponding required values. Inaddition, a collaborative optimization technique with several objectives is used to adjust the regulatorparameters. The capability and efficiency of the suggested regulator is first investigated throughcomputer simulation results and then confirmed by real-time control experimental results on the QTSbased on dSPACE 1104 computation engine. The findings showed that the suggested OH-FO-T2F-PIDregulator significantly outperformed both the optimized ADRC and the OH-FO-T1F-PID regulators.Specifically, it reduced the rising time by 17.02% and 95.21%, respectively, and the settling time by25.13% and 74.28%. Additionally, the designed OH-FO-T2F-PID regulator successfully eliminatedthe steady-state error and overshoot, enabling precise regulation of the QTS, and maintenance theliquid level at the desired set point under a wide range of working situations. The robustness of therecommended regulator is also studied by considering − 50% disturbance in the QTS parameters, andthe findings showed that the OH-FO-T2F-PID regulator is less susceptible to variations in parameters.Keywords: Quadruple tank system (QTS), Optimal hybrid fractional order type 2 fuzzy PID regulator,Hybrid A-GWOCSO algorithm, Multi-objective optimization, dSPACE 1104 computation engine

Composite insulators demonstrate superior electrical performance in contrast to standard insulators.Nevertheless, the deterioration of composite insulator and the challenges in identifying defects arethe primary drawbacks of these insulators. This study investigates the effect of water droplets onthe electrical behavior of composite insulators, which are widely used in high-voltage applications.Using COMSOL software, a Finite Element Model (FEM) was developed to simulate the electric fielddistribution on the surface of a composite insulator in the presence of water droplets. The resultsindicate that the existence of water droplets increases the electric field intensity by approximately33.33% when the number of droplets increases from two to six. The simulations also reveal that waterdroplets significantly increase the electric field’s intensity, which affects the electric field and potentialdistribution on the insulator’s surface. Furthermore, the conductivity of water droplets was found tohave a negligible impact on the electric field distribution along the insulator. To systematically evaluatethe influence of various factors, Response Surface Methodology (RSM) was employed in combinationwith Analysis of Variance (ANOVA) to analyze the interactions between water droplet number,pollution, and applied voltage. The statistical analysis demonstrated that the maximum electric fieldintensity increased by nearly 38.3% as water droplet conductivity rose from low to high levels. RSMwas used to generate a second-order polynomial model that describes the relationship between thesefactors and the electrical performance of the insulator, allowing for the identification of significanttrends and interactions. The findings provide valuable insights for the design and development ofcomposite insulators that are more resilient to environmental factors, enhancing their overall electricalperformance.Keywords: ANOVA, Composite insulator, Electric field, FEM, Water droplet