DC-DC buck converters are inherently nonlinear systems that often operate under dynamically changing conditions, parameter uncertainties, and external disturbances, posing significant challenges for conventional control strategies. This paper introduces a novel cascaded proportional-integral and proportional-derivative (PI-PD) controller architecture, in which all controller parameters are optimally tuned using the recently developed Electric Eel Foraging Optimizer (EEFO), a bio-inspired metaheuristic algorithm modeled on the electrolocation and hunting behaviors of electric eels. The proposed control structure uniquely integrates a dual-loop configuration: the inner PI loop eliminates steady-state error, while the outer PD loop enhances dynamic response and mitigates rapid transient fluctuations. This cascaded arrangement enables decoupled tuning of steady-state and transient characteristics, offering superior control flexibility compared to conventional single-loop PID designs. To calibrate the controller, EEFO is employed to minimize a composite performance objective function that simultaneously considers settling time and overshoot, ensuring well-damped and rapid system behavior. A comprehensive set of simulation experiments was conducted in a MATLAB/Simulink environment to evaluate the proposed method against multiple benchmark algorithms-including the flood algorithm, gazelle optimization algorithm, and artificial hummingbird algorithm-as well as classical PID, PID acceleration (PIDA), and fractional-order PID (FOPID) controllers optimized by state-of-the-art metaheuristics. Across all key performance metrics-including rise time, settling time, percentage overshoot, peak time, and steady-state error-the EEFO-tuned cascaded PI-PD controller demonstrated consistently superior results, achieving near-zero overshoot, ultra-fast convergence, and minimal output deviation. Beyond nominal conditions, extensive robustness analyses were conducted to validate the controller's effectiveness under realistic disturbances, such as abrupt load changes, high-frequency measurement noise, time-delay effects in feedback channels, and +/- 10%-15% parametric variations in inductance and capacitance. In all scenarios, the controller retained stable output regulation, confirming its resilience and practical viability. To the best of our knowledge, this is the first study to deploy a cascaded PI-PD control structure specifically designed for DC-DC buck converters and optimized using the EEFO algorithm. The integration of a biologically inspired optimization framework with a decoupled dual-loop control scheme offers both architectural and algorithmic novelty. The proposed strategy addresses critical demands in nonlinear converter regulation and provides a robust, high-performance solution suitable for dynamic and uncertain power electronic environments.

Acoustic signal (AS) has emerged as a powerful non-contact technique for early detection of incipient faults in aircraft rotating components due to its high sensitivity to transient damage mechanisms. However, the strong non-stationarity and noise susceptibility of acoustic signals under dynamically varying operating conditions present significant challenges for reliable fault diagnosis. In this study, an explainable deep learning framework based on a Long Short-Term Memory (LSTM) network optimized with the AdamW algorithm is proposed for fault diagnosis of aircraft-related rotating components using acoustic signals. The framework leverages sequential learning to capture the temporal evolution of acoustic signals and is systematically compared with conventional recurrent architectures, including Recurrent Neural Networks (RNNs) and Gated Recurrent Units (GRUs). Experimental results demonstrate that the proposed LSTM–AdamW model achieves superior diagnostic performance, reaching a test accuracy and macro-F1 score of 99.26% under dynamic operating conditions. The enhanced performance is attributed to the LSTM’s ability to model long-term temporal dependencies and the regularization benefits of the AdamW optimizer through decoupled weight decay. To improve transparency and physical interpretability, explainable artificial intelligence techniques based on Local Interpretable Model-Agnostic Explanations (LIME) and SHapley Additive exPlanations (SHAP) are employed. The explainability analysis reveals that classification decisions are driven by localized, physically meaningful transient acoustic patterns associated with fault-induced events. In addition, a Taylor diagram–based statistical assessment confirms strong agreement between model predictions and reference signals, indicating robust preservation of temporal signal characteristics. The results suggest that the proposed explainable LSTM–AdamW framework provides a reliable, computationally efficient, and interpretable solution for acoustic signal-based fault diagnosis in aerospace applications, with strong potential for real-time condition monitoring and predictive maintenance.

As decentralized energy systems gain momentum, microgrids (MGs) have become a vital component of the modern power landscape. Yet, maintaining power quality (PQ) within these systems presents ongoing challenges due to the presence of nonlinear loads, variable renewable energy sources, and frequent switching operations. These factors contribute to PQ disturbances, such as harmonic distortion, voltage instability, and synchronization issues. Conventional mitigation methods often struggle to cope with such dynamic and complex environments. This review investigates the emerging role of artificial intelligence (AI) as a powerful tool for optimizing PQ in MGs. It presents a detailed overview of various AI-based methods, including machine learning (ML), metaheuristics, deep learning, fuzzy logic, and hybrid approaches and their implementation in areas like harmonic suppression, voltage and frequency regulation, islanding detection, renewable energy coordination, and predictive diagnostics. The study evaluates these techniques based on key performance indicators, such as precision, scalability, and suitability for real-time operation, while also addressing challenges related to data reliability, interpretability, and cybersecurity. The article concludes by highlighting future research directions, such as AI integration with Internet of Things (IoT), edge computing, and decentralized intelligence. Overall, the review illustrates how AI can play a pivotal role in transforming MG PQ optimization for the evolving smart grid era.

In petahertz (PHz)-based optical wireless communication (OWC), being proposed as an unbeatable wireless link for underwater communication, the channel model is characterized by both line-of-sight (LOS) and non-LOS (NLOS) components. On the other hand, additive white Gaussian noise (AWGN) distribution fails to appropriately characterize noise in an underwater scenario. However, recently proposed α-Beaulieu-Xie fading distribution, with α as a modified parameter, includes both NLOS and LOS components simultaneously to characterize wireless channel in an underwater scenario. Also, additive white generalized Gaussian noise (AWGGN), instead of AWGN distribution, shows perfect match with experimental data. In this research, performance analysis over α-Beaulieu-Xie fading in presence of AWGGN noise is presented under dual selection combining (SC) diversity. In particular, analytical expressions for error rate (ER) of modulation schemes such as BPSK, MQAM, and MPSK modulations over α-Beaulieu-Xie fading in presence of AWGGN noise are presented under dual SC diversity. In addition, asymptotic expressions of respective ER are presented. Effect of various fading parameters such as overall severity parameter of both LoS and NLoS (mX), severity parameter of LoS (mY), and fading parameter (α) on the presented error rates is quantified numerically. Also, the effect of noise parameter ‘p’ on ER of different modulation schemes is analyzed. Also, presented results are compared with those of closely related results available in the literature. This also proves the validity of presented expressions.

This paper proposes a novel intelligent control framework for precise speed regulation of DC motors using a two-degree-of-freedom proportional-integral-derivative (2-DOF PID) controller whose six parameters are optimally tuned via the animated oat optimization (AOO) algorithm. AOO is a nature-inspired metaheuristic derived from the hygroscopic seed dispersal behavior of Avena sterilis L., offering a hybrid global-local search strategy that balances exploration and exploitation through mechanisms such as eccentric rotation, projectile jumps, and L & eacute;vy flights. The controller parameters are optimized within bounded physical limits using a composite cost function that integrates Integral Absolute Error (IAE) and normalized overshoot percentage. Comparative simulations are conducted against four recent metaheuristics, catch fish optimization algorithm (CFOA), greater cane rat algorithm (GCRA), RIME optimization, and particle swarm optimization (PSO), under identical settings (100 iterations, 30 agents). Statistical analysis over 25 independent runs reveals that the AOO algorithm achieves the best average cost value (65.2185) with the lowest standard deviation (1.2091), indicating high convergence stability. In time-domain analysis, the AOO-based 2-DOF PID controller exhibits the shortest rise time (0.9217 s), fastest settling time (1.4143 s), lowest overshoot (0.5688%), and negligible steady-state error (similar to 10-13), outperforming all benchmark algorithms. The controller's robustness and adaptability are further validated through simulations under random reference inputs, multistep transitions, and external disturbances with measurement noise. Experimental validation using a real-time hardware-in-the-loop setup, including a Pololu 12 V DC geared motor, IBT-2 H-bridge driver, Arduino Uno, and MATLAB/Simulink interface, confirms the simulation results, demonstrating fast, stable, and accurate speed tracking across diverse reference profiles. These findings confirm that the AOO-based 2-DOF PID controller offers a highly effective, reliable, and practically deployable solution for high-performance motor control under dynamic and uncertain operating conditions.

In this article, we have studied the double-star permanent magnet synchronous machine, and two mathematical models have been proposed in Park reference frame: the d1q1d2q2O1O2 model and the dqz1z2O1O2 model. This machine is controlled by vector control on both models. This is followed by the estimation of the speed and the position of the machine by two techniques; the first is the estimation of the back EMF of the model using fuzzy logic and then the determination of the speed and the position using PLL. And the second is model reference adaptive system (MRAS) observer speed of the machine. The simulation of the sensorless control of the machine is carried out under the MATLAB/Simulink software environment and gives encouraging results.

The paper presented study dealing with corona discharge field of an ultra-high voltage transmission line. Methods determination of corona power losses are highlighted in the research. Proposed method for defining of corona power losses on transmission power line wires has been presented, which, unlike the existing ones, allows measuring corona power losses directly.

Bearing degradation is the primary cause of electrical machine failures, making reliable condition monitoring essential to prevent breakdowns. This paper presents a novel hybrid model for the detection of multiple faults in bearings, combining Long Short-Term Memory (LSTM) networks with random forest (RF) classifiers, further enhanced by the Grey Wolf Optimization (GWO) algorithm. The proposed approach is structured in three stages: first, time and frequency domain features are manually extracted from vibration signals; second, these features are processed by a dual-layer LSTM network, which is specifically designed to capture complex temporal relationships within the data; finally, the GWO algorithm is employed to optimize feature selection from the LSTM outputs, feeding the most relevant features into the RF classifier for fault classification. The model was rigorously evaluated using a dataset comprising six distinct bearing health conditions: healthy, outer race fault, ball fault, inner race fault, compounded fault, and generalized degradation. The hybrid LSTM-RF-GWO model achieved a remarkable classification accuracy of 98.97%, significantly outperforming standalone models such as LSTM (93.56%) and RF (98.44%). Furthermore, the inclusion of GWO led to an additional accuracy improvement of 0.39% compared to the hybrid LSTM-RF model without optimization. Other performance metrics, including precision, kappa coefficient, false negative rate (FNR), and false positive rate (FPR), were also improved, with precision reaching 99.28% and the kappa coefficient achieving 99.13%. The FNR and FPR were reduced to 0.0071 and 0.0015, respectively, underscoring the model’s effectiveness in minimizing misclassifications. The experimental results demonstrate that the proposed hybrid LSTM-RF-GWO framework not only enhances fault detection accuracy but also provides a robust solution for distinguishing between closely related fault conditions, making it a valuable tool for predictive maintenance in industrial applications.

This paper explores scenarios for powering rural areas in Gaita Selassie with renewable energy plants, aiming to reduce system costs by optimizing component numbers to meet energy demands. Various scenarios, such as combining solar photovoltaic (PV) with pumped hydro-energy storage (PHES), utilizing wind energy with PHES, and integrating a hybrid system of PV, wind, and PHES, have been evaluated based on diverse criteria, encompassing financial aspects and reliability. To achieve the results, meta-heuristics such as the Multiobjective Gray wolf optimization algorithm (MOGWO) and Multiobjective Grasshopper optimization algorithm (MOGOA) were applied using MATLAB software. Moreover, optimal component sizing has been investigated utilizing real-time assessment data and meteorological data from Gaita Sillasie, Ethiopia. Metaheuristic optimization techniques were employed to pinpoint the most favorable loss of power supply probability (LPSP) with the least cost of energy (COE) and total life cycle cost (TLCC) for the hybrid system, all while meeting operational requirements in various scenarios. The Multi-Objective Grey Wolf Optimization (MOGWO) technique outperformed the Multi-Objective Grasshopper Optimization Algorithm (MOGOA) in optimizing the problem, as suggested by the results. Furthermore, based on MOGWO findings, the hybrid solar PV-Wind-PHES system demonstrated the lowest COE (0.126€/kWh) and TLCC (€6,897,300), along with optimal satisfaction of the village's energy demand and LPSP value. In the PV-Wind-PHSS scenario, the TLCC and COE are 38%, 18%, 2%, and 1.5% lower than those for the Wind-PHS and PV-PHSS scenarios at LPSP 0%, according to MOGWO results. Overall, this research contributes valuable insights into the design and implementation of sustainable energy solutions for remote communities, paving the way for enhanced energy access and environmental sustainability.

A reliable autonomous navigation system in environments with complete darkness, fog, and smoke is still an open problem in the field of mobile robots, where traditional computer vision approaches are not applicable because they heavily rely on the illumination and sightline conditions. This paper proposes an embedded hardware framework based on the ROS platform that provides reliable environmental perception and decision-making using the fusion of the thermal and LiDAR sensors, along with the machine learning-based perception module. The proposed framework is implemented using the Raspberry Pi 4 platform to show the possibility of using such advanced multi-sensor fusion techniques on embedded hardware devices. The embedded hardware framework provides a spatiotemporal environmental model using the geometric information obtained from the 2D LiDAR sensor and the thermal intensity information. Real-time obstacle detection and classification are carried out using a YOLOv5-based CNN on the fused stream of data, ensuring accurate detection even under poor lighting conditions. Experimental evaluation of the proposed approach under complete darkness, smoke, and fog showed 96%-97% accuracy, achieving better performance than thermal-only and LiDAR-only approaches by 12% and 8% respectively. The proposed fused framework is found to be feasible for real-time operation on Raspberry Pi 4 devices with a 100% success rate.

As the solar photovoltaic (PV) penetration level increases in smart grids, precise and computationally efficient short-term forecasting becomes essential to aid operational planning and real-time energy management. However, the power produced by PV is highly nonlinear and stochastic due to variations in weather factors, which weakens the performance of single forecasting models. The aim of this work is to propose a stacked ensemble regression model that combines Gradient Boosting and XGBoost (Extreme Gradient Boosting) as base learners, with Ridge Regression as the meta-learner, for very short-term PV power prediction. The model operates using meteorological and operational parameters, such as temperature, humidity, wind profile, cloudiness distribution, and solar situation. Standard preprocessing steps (missing value imputation, feature selection, and normalization) are adopted to facilitate stable model training. An empirical study is carried out using real-world PV generation data, and the results are compared with popular gradient boosting algorithms such as Gradient Boosting, XGBoost, LightGBM (Light Gradient Boosting Machine), and CatBoost (Categorical Boosting), and machine learning models such as multilayer perceptron (MLP) and LSTM, using k-fold cross-validation. The boosted ensemble improves predictive accuracy, achieving MAE = 0.042 ± 0.002, MSE = 0.0031 ± 0.0002, and R² = 94% ± 1% under the experimental conditions used in this work. Nonparametric tests (i.e., the Wilcoxon signed-rank test and the Friedman test) show that such improvements are statistically significant (p < 0.05). Moreover, the inference latency of the proposed model is quite low, which demonstrates its suitability for near-real-time deployment in real-world smart grid scenarios. According to experimental results, lightweight ensemble learning can stand as a competitive and practical alternative to complicated deep learning methods for short-term PV power forecasting when data availability and computational budget are taken into account.

Surface roughness in CNC turning is a pivotal quality metric shaping functional performance, service life and production cost. This study investigates data-driven prediction of arithmetic mean surface roughness (Ra) during the turning of AISI H13 steel under both new-tool and progressively worn-tool conditions. Several machine learning models including k-Nearest neighbors (KNN), random forest (RF) and extra trees (ExT) are evaluated and compared with a stacking ensemble model that integrates these base learners using a linear regression meta-learner. The input variables consist of cutting speed, feed rate, depth of cut and triaxial cutting force components. The results show that the KNN model exhibits limited predictive accuracy whereas the RF and ExT models achieve competitive performance. The proposed stacking ensemble consistently outperforms all individual models achieving a coefficient of determination (R²) exceeding 0.98 along with substantial reductions in root mean square error (RMSE) and mean absolute error (MAE) under tool-wear conditions indicating strong generalization capability. To enhance model transparency SHapley additive exPlanations (SHAP) and local interpretable model-agnostic explanations (LIME) are employed. The interpretability analyses identify feed rate as the dominant factor influencing surface roughness while the importance of cutting forces and the interaction between depth of cut and feed rate increases as tool wear progresses. Overall the findings demonstrate that the proposed stacking-based hybrid model provides an accurate, robust and explainable framework for surface roughness prediction in CNC turning offering practical potential for in-process quality monitoring and decision support applications.

Effective performance evaluation and fault detection in PV strings rely heavily on current-voltage (I-V) curve analysis. Yet, most commercial tracers are prohibitively expensive, interrupt system operation during testing and lack the capability for real-time monitoring in high-voltage, high-power environments. To overcome these challenges, this study presents a low-cost, real-time, and wireless solution-an Online I-V Tracing Tool (OIVTT)-engineered for high-voltage photovoltaic strings deployed in utility-scale solar power systems. The OIVTT employs a dual-branch Insulated Gate Bipolar Transistor (IGBT)-based electronic load driven by an STM32 microcontroller, enabling dynamic I-V curve tracing with precise control across the full operating range. The system is capable of tracing I-V curves for PV strings up to 1500-V open-circuit voltage and 20-A short-circuit current and performs complete curve acquisition under one second, ensuring compliance with the IEC 60904 standard for solar measurements under stable irradiance and temperature conditions. A wireless communication framework, developed using the Kodular mobile application platform, allows remote initiation and control of measurements, with real-time data synchronized to the Firebase cloud. For enhanced sampling accuracy, the system employs an adaptive dual-branch control algorithm that continuously modifies sweep parameters in response to the I-V curve's gradient, thereby ensuring balanced placement of data points across its steep and flat regions. The algorithm maintains a relative standard deviation (RSD) below 10%, indicating high consistency and precision of data acquisition. The system was experimentally validated using a 15-module monocrystalline PV string with a total capacity of 3.6 kW, tested under diverse scenarios such as consistent sunlight exposure, induced partial shading, and deliberate modifications to series and shunt resistances. The OIVTT successfully captured performance anomalies and matched theoretical expectations, demonstrating an average deviation of less than 2% when compared with the standard Shockley diode-based PV model. Under partial shading, the system accurately detected bypass activation events and the presence of stepped I-V profiles, while tests on altered series/shunt resistance confirmed sensitivity to fill factor degradation. In comparison with commercial I-V tracers-which typically require offline testing, exhibit longer sweep durations (2-9 s), and cost Indian Rupee12L-Indian Rupee20L-the proposed OIVTT provides a high-speed, accurate, and economically viable solution (< Indian Rupee35k total cost). Its nonintrusive operation, wireless control, and high-voltage support make it ideal for continuous monitoring, preventive diagnostics, and maintenance in utility-scale PV systems.

Commutation failures (CFs), which occur when current transfer between valves in line-commutated converter high-voltage direct current (LCC-HVDC) systems is disrupted, pose a challenge in weak alternating current (AC) networks. This paper introduces a coordinated control strategy that combines a fuzzy self-tuning proportional-integral (PI) controller (FSTPIC) and a static synchronous compensator (STATCOM) device to mitigate CFs and enhance system stability. The approach applies the FSTPIC to both converters of the HVDC link, while the STATCOM at the inverter side delivers dynamic reactive power and voltage support during AC faults. We test this strategy on the CIGRE HVDC benchmark system using MATLAB/SIMULINK simulations. The results demonstrate that the proposed method significantly reduces CFs, mitigates transient oscillations, and shortens recovery time compared to conventional control techniques. This coordinated control boosts voltage stability and the system’s ability to ride through faults, confirming its superiority under various fault scenarios in weak-grid conditions.

Accurate and robust sensorless speed control of Permanent Magnet Synchronous Motor (PMSM) drives is essential for high-performance Electric Vehicle (EV) applications operating under rapid speed variations, load disturbances, and repeated start–stop conditions. This paper proposes a robust sensorless PMSM control framework based on Integral Sliding Mode Control (ISMC) combined with Extended Kalman Filter (EKF) and Model Reference Adaptive System (MRAS) observers. Unlike previous studies focusing separately on control or observer design, this work provides an integrated robustness and estimation accuracy evaluation under realistic EV driving conditions. The proposed ISMC scheme is designed to improve disturbance rejection capability and eliminate the conventional reaching phase of sliding mode control. EKF and MRAS observers are employed for sensorless rotor speed estimation, and their performances are comparatively analyzed under multiple EV operating scenarios, including step speed changes, dynamic driving profiles, and repeated start–stop operation. In addition, a weighted performance index is introduced to quantitatively compare the considered control configurations. Simulation results demonstrate that the proposed ISMC–EKF configuration achieves superior transient and steady-state performance compared with conventional PID-based control. The rise time is reduced from approximately 3.5 ms to 2.7 ms, while the settling time decreases from nearly 10 ms to about 5–6 ms. Furthermore, overshoot is reduced from 5% to 0.3%, and torque ripple is significantly minimized under dynamic EV conditions. Robustness analysis under parameter variations additionally confirms the disturbance rejection capability and stability of the proposed framework. The obtained results demonstrate that the proposed ISMC–EKF framework provides an efficient and reliable solution for high-performance sensorless PMSM drives in next-generation EV propulsion systems.

The increasing proliferation of smart grids and the growing share of renewable energy sources call for innovative and intelligent approaches to energy distribution management. Conventional energy management techniques encounter significant limitations, including suboptimal energy allocation, elevated operational expenses, and limited adaptability to dynamic load variations within the network. This study introduces an advanced smart grid architecture that incorporates IoT-based sensors and a control mechanism powered by deep learning algorithms. By leveraging data from IoT devices and centralized databases, the proposed system enables continuous monitoring of grid parameters, supports real-time analytics, and facilitates adaptive and predictive decision-making. These capabilities contribute to enhanced energy distribution efficiency, reduced technical losses, and improved overall system reliability. Furthermore, the architecture ensures robust resource allocation, even under conditions of unforeseen failures of energy assets, including generation units, distribution infrastructure, or end-users. The system also supports accurate demand forecasting and contributes to maintaining grid stability. Through the integration of IoT technologies, deep learning models, and real-time data processing, the proposed intelligent energy management framework is well equipped to address the challenges of increasing energy demand and the variability inherent in renewable energy generation.

This paper provides an experimental evaluation of two current transformers' (CTs') accuracy performance under linear and nonlinear load scenarios. Four performance indicators-measured RMS current, total harmonic distortion (THD), ratio error, and phase error-were assessed across seven sample load scenarios in compliance with IEEE C57.13 and IEC 61,869 standards. The findings demonstrate that whereas CT accuracy is high under near-sinusoidal stimulation, it deteriorates with increasing waveform distortion and load current. Higher current levels exacerbate core nonlinearity and saturation, resulting in larger ratio and phase errors, while increased harmonic content increases THD and causes phase displacement between primary and secondary currents. In nonlinear circumstances, Actual current magnitudes are underestimated due to the constant attenuation of observed RMS values. This work's main contribution is a synchronized, time-domain experimental evaluation showing that phase error changes in tandem with ratio error and THD as functions of current magnitude and harmonic content. The limitations of sinusoidal-based CT accuracy classifications are revealed by this experimentally verified coupling between amplitude and angular inaccuracies. It also emphasizes the potential of sophisticated corrective techniques, such as real-time error detection, harmonic compensation, and artificial intelligence-based predictive modeling, to enhance CT reliability in distortion-rich power networks.

This research addresses the challenge of accurately measuring total harmonic distortion (THD) in smart meters, where traditional methods, such as FFT-based estimation, often fail to maintain high accuracy, especially when using low-cost sensors. The primary objective is to evaluate the potential of artificial intelligence (AI) techniques to improve THD estimation despite the limitations of economic sensor technology. To achieve this, the study employs a range of AI models, including AdaBoost, Random Forest, Support Vector Machines (SVM), and Artificial Neural Networks (ANN), utilizing data initially gathered from high-resolution sensors, which provide accurate reference measurements. Lower-resolution data were then extracted to simulate the performance of economic sensors, which typically offer less precise measurements. Among the AI models tested, AdaBoost consistently outperformed the others in terms of accuracy. In contrast, traditional FFT methods exhibited substantial performance degradation, particularly with low-cost sensors. By applying artificial intelligence, a trained model was developed, where inputs from economic sensors are processed to yield outputs that closely approximate those of high-precision sensors. The results confirm that integrating AI algorithms, especially AdaBoost, improves the performance of low-cost sensors, enabling them to achieve results comparable to high-precision, expensive sensors. This breakthrough not only provides a scalable and cost-effective solution for power quality analysis but also facilitates the development of affordable, high-quality sensors that can be deployed in smart meters without compromising measurement accuracy. The results were also experimentally confirmed in real-world scenarios, further validating the effectiveness of the AI-enhanced approach.

This paper presents an advanced predictive current control strategy for voltage source inverters (VSIs) based on an artificial neural network (ANN)-enhanced deadbeat (DB) controller. Unlike conventional DB controllers, whose performance strongly depends on accurate knowledge of the filter resistance and inductance, the proposed ANN-DB controller executes in real time without requiring explicit system parameters or analytical models, relying solely on measured currents and voltages. The ANN is trained offline using data generated from a traditional DB current controller over a wide range of operating conditions, and then deployed as a data-driven approximation of the deadbeat control law. Experimental results on a SEMIKRON two-level VSI with an RL load, implemented on a dSPACEMicroLabBox platform, demonstrate that the ANN-DB controller achieves faster transient response (settling time of approximately 1.4 ms), reduced current total harmonic distortion (THD approximate to 1.9%), and improved steady-state tracking accuracy compared with the conventional DB approach under varying load conditions. Robustness tests with inductance variations from 0.09 H to 0.18 H confirm that the ANN-DB controller maintains low RMS current error and low THD, whereas the conventional DB controller exhibits pronounced sensitivity to parameter deviations. A comparative discussion with PI, model predictive control (MPC), classical DB, and recent model-free predictive current control methods highlights that the proposed ANNDB scheme offers a favourable compromise between dynamic performance, robustness, and computational complexity for advanced power electronic applications.

Prompt and accurate fault detection in extra high voltage transmission lines is required for guaranteeing the steadiness of power system. This study describes the performance of BiLSTM, GRU, and TCN as deep learning models for the detection and classification of faults in transmission lines through synthetic and real-time sequential datasets in 500 kV transmission line between Jamshoro and Karachi (NKI), in Sindh, Pakistan. Testing models' performance on simulated faults versus real fault events, the study concludes a major space and suggests insights for their practical applicability. The results show that deep learning models can reach vast level of accuracy in classifying different faults in transmission lines. This study forms the basis for exploiting modern fault detection practices in operating grids to improve their dependability and flexibility. The results revealed an accuracy of 98.31%, achieved by the BiLSTM, 94.27% for GRU and TCN as 99.8% through simulated data set, whereas using real-time fault data BiLSTM scored 62.05% accuracy, while GRU accuracy score achieved 96.43%, and TCN attained 100% accuracy. The results demonstrate that the deep learning models used in this study work well analyzing time series data by achieving high fault accuracy for fault classification in transmission lines. In general, the study was conducted to identify the best model in managing the fault over extra high voltage transmission lines under different conditions.