Azzeddine Dekhane

Artificial intelligence (AI)-based control is gaining increasing attention in electrical-machine drives as a means of approximating nonlinear control laws through data-driven learning. Neural behavioral cloning provides a practical framework for reproducing an existing analytical controller while retaining the established control architecture. However, neural controllers are typically selected according to supervised validation loss, despite the fact that its ability to predict closed-loop control performance remains unclear. This paper systematically investigates the relationship between supervised learning accuracy and closed-loop performance through neural behavioral cloning of an integral sliding-mode speed controller for a 2.2 kW field-oriented induction motor drive. A bias-free, odd-symmetric neural network is trained to approximate the analytical control law, while the integral state, state-update law, anti-windup mechanism, actuator saturation, execution sequence, and controller lifecycle are kept identical in the analytical and neural implementations, thereby isolating the effect of the learned approximation on closed-loop behavior. The controllers are evaluated under ten benchmark conditions comprising nominal operation, six individual parameter perturbations, and a combined worst-case condition, with the sliding-surface coefficient selected equal to the PI design bandwidth parameter. Across five trained networks, whose validation mean-squared errors span a factor of 2.4, supervised validation accuracy exhibits only a weak relationship with closed-loop performance. The largest ITAE variation among the five networks within any benchmark cell is 1.60% of the corresponding cell median, while the network with the highest validation error yields the lowest ITAE in nine of the ten benchmark cells. Nevertheless, all neural controllers closely reproduce the analytical controller and achieve a 39–68% reduction in ITAE relative to a conventional proportional–integral controller, with a median reduction of 49% across the ten benchmark cells. These results indicate that, within the investigated setting, supervised validation loss is not a reliable criterion for ranking behaviorally cloned controllers according to their closed-loop performance. The findings are limited to a single induction motor drive, one teacher controller, one network architecture (H = 16), and five trained networks representing two distinct validation-error groups; broader conclusions require systematic evaluation across architectures, teacher controllers, drive conditions, and training configurations.

The rapid expansion of photovoltaic (PV) installations necessitates advanced monitoring solutions to ensure optimal performance, early fault detection, and long-term reliability. This study presents a hybrid methodology for real-time PV module monitoring and fault diagnosis, integrating I–V curve analysis, IEC 60891 standardization, and Teaching–Learning-Based Optimization (TLBO) to estimate parameters of the Single Diode Model (SDM). An I-V Swinger-based test bench equipped with Arduino data acquisition and environmental sensors was developed to measure I–V characteristics under various operating conditions (OPC), including low irradiance, elevated temperature, and partial shading. The IEC 60891 procedure enabled OPC-to-STC conversion, facilitating accurate performance evaluation and degradation rate estimation. Concurrently, the TLBO algorithm efficiently extracted SDM parameters (I_L, I_o, R_s, R_sh, a), achieving a mean absolute percentage error (MAPE) below 5% and an R² of 0.99, confirming the method’s robustness. Experimental results on an 11-year-old LX-80M module revealed moderate degradation (<13%) and demonstrated the system’s ability to capture shading-induced power losses and fill factor collapse, distinguishing between permanent aging and transient faults. A comparative analysis with the recent literature confirms that this integrated, low-cost, and high-accuracy approach bridges the gap between academic PV diagnostics and industrial monitoring needs. The proposed framework provides a foundation on which IoT-enabled predictive maintenance and digital-twin integration could subsequently be built; neither extension was implemented or validated in the present study.

This paper investigates the effectiveness of integrating a matrix converter (MC) into a wind turbine system based on a Doubly Fed Induction Generator (DFIG), focusing on the indirect control of active and reactive power using fractional regulators. To evaluate the performance of this architecture, three distinct configurations are studied and compared: (i) a conventional system using an AC-DC-AC converter combined with a conventional PI regulator, (ii) a structure integrating a matrix converter with a PI regulator, and (iii) an advanced combination of MC with a fractional regulator. Simulation results demonstrate that this latter solution offers significant improvements in dynamic performance, more precise regulation of active and reactive power, and higher energy efficiency than the other two methods.

This study presents a novel approach to optimizing water purification systems at the Zaouiet Kounta solar power plant through the integration of advanced automation and supervision technologies. By utilizing a siemens programmable logic controller (PLC) and human-machine interface (HMI) programmed via the totally integrated automation (TIA) Portal software, the project aimed to significantly enhance the performance of water production and distribution systems. The objectives included improving operational efficiency, reducing manual intervention, and increasing system reliability and precision. The results presented herein show significant improvements in operational efficiency, system reliability, and automation in a challenging environmental context. This research provides a comprehensive case study that not only highlights the feasibility of using Siemens PLC and HMI systems in solar-powered water purification systems but also proposes scalable solutions for similar industrial applications.

This work presents a new integration of the Whale Optimization Algorithm (WOA) with Finite Control Set Model Predictive Control (FCS‑MPC) to optimise Fractional‑Order Proportional–Integral (FOPI) controllers for Multilevel Inverters (MLIs)—a combination hitherto unexplored in the literature. In the proposed hierarchical scheme, the WOA‑tuned FOPI operates in the outer loop to generate highly accurate and adaptive current reference signals, while FCS‑MPC forms the inner loop, predicting the inverter’s behaviour for all possible switching states and selecting the one that minimises a defined cost function. This coordinated outer–inner control design enhances reference tracking accuracy and reduces steady‑state error, leading directly to improved harmonic suppression, voltage regulation, and dynamic stability. The proposed WOA‑FOPI‑FCS‑MPC control achieves a 12–15% reduction in Total Harmonic Distortion (THD) and a 22% improvement in dynamic stability compared to conventional PI‑based predictive control. Additionally, a systematic comparison between diode‑clamped and T‑type MLIs in a 1 MW PV‑grid system reveals that the T‑type inverter reaches 98.5% efficiency and 1.48% voltage THD at 7‑level operation, offering an optimal trade‑off between switching device count, power quality, and efficiency. Statistical analysis—including mean, minimum, maximum, standard deviation, and computational time—confirms the robustness and consistency of the proposed optimisation, while benchmark function evaluations validate the global search capability of WOA. The results, validated through MATLAB/Simulink simulations and preliminary experimental tests, demonstrate the method’s strong potential for high performance industrial PV grid integration.

The efficacy of photovoltaic (PV) systems significantly depends on optimizing power extraction under diverse environmental circumstances. The examination of Maximum PowerPoint Tracking (MPPT) approaches, particularly those utilizing soft computing (SC), is essential. They provide adaptability and accuracy for nonlinear situations such as partial shade and swift variations in irradiance. This article evaluates and contrasts SC-based MPPT methodologies, encompassing Fuzzy Logic Control (FLC), Artificial Neural Networks (ANN), Adaptive Neuro-Fuzzy Inference Systems (ANFIS), and Evolutionary Algorithms (EA), emphasizing their distinct characteristics, benefits, and drawbacks.