Rachid Bendaoud

Accurate modeling of power electronic converters under parametric uncertainty remains a challenge for purely data-driven approaches. Although physics-informed neural networks (PINNs) improve generalization by embedding physical constraints, many existing methods rely on ideal lossless assumptions and enforce only algebraic relations such as power balance. This paper proposes a residual and loss-aware physics-informed neural network framework for robust modeling of non-ideal boost converters. The approach introduces two key features: (i) dynamic residual constraints derived from the large-signal averaged state-space equations of the converter, and (ii) explicit incorporation of non-ideal loss mechanisms, including inductor copper losses, within the physics-informed formulation. By enforcing differential dynamics together with realistic energy dissipation, the proposed model improves physical consistency under practical operating conditions. A comprehensive simulation dataset is generated across wide parameter ranges, including variations in inductance, capacitance, load resistance, and input voltage. To evaluate robustness, structured out-of-distribution tests are performed using inductance values outside the training range. Comparative results show that the proposed residual and loss-aware PINN significantly reduces prediction error, improves dynamic consistency, and enhances generalization compared with conventional neural networks and power-balance PINNs. These results highlight the importance of differential residual enforcement and non-ideal modeling in physics-informed learning for power electronic systems.

An increase in operating temperature of the crystalline photovoltaic modules (PVMs) has a significant impact on their electrical efficiency. Accurate determination of this parameter is crucial for a credible evaluation of PVM performance. Due to the difficulty of analytically solving the energy balance equations (EBEs) to determine the temperature profile of the various layers of the PVM, numerical solutions are often chosen. The exact analytical solution of these equations is challenging to achieve due to the complexity of some time‐dependent parameters (solar irradiance, ambient temperature, wind speed, front and rear surface temperatures, …) and can only be reached in particular cases. To overcome this difficulty, some approximations are considered and permit the resolution of this problem analytically. In this study, a one‐dimensional model is developed. It is based on an analytical solution of the steady‐state EBEs. The analytical solution is reached by assuming that the time‐varying parameters can be considered constant over more or less long time intervals, depending on the change rate of these parameters. The objective of this study is to propose an analytical model improving the prediction accuracy of the temporal and spatial evolution of the PVM operating temperature compared with literature results; meanwhile, simplifying the estimation process of the various photovoltaic system parameters. The experimental and modeling results obtained are compared with those evaluated using other relevant models. The compatibility between the experimental and model results attests to the reliability of the assumed approximations, with regard to calculated RMSE (0.78) and MBE (0.77) values.

Photovoltaic (PV) technology is generally perceived as well‐developed but suffers a drop in performance at high temperatures. Faced with this problem, researchers are turning to PV thermal (PVT) systems, which integrate electricity production and thermal energy. Flat‐plate PVT systems are the most widely adopted among the various configurations. This article is distinguished by an in‐depth analysis of flat‐plate PVT systems, drawing on a detailed analysis of recent research. It summarizes the numerous studies on the different layers of PVT systems, providing an overview of advances in this field. The materials used for absorbers and tubes are explored, providing information on their properties and applications and on the research being carried out to optimize their efficiency. The analysis also focuses on heat exchanger, tube, and channel configurations, highlighting innovations to improve their performance. Methods for integrating absorbers and tubes with PV panels, the most efficient types of PV cells, and working fluids for optimizing heat transfer and thermal performance are also discussed. Finally, an overview of software tools for simulating PVT systems and a summary of research on each software tool are provided to help researchers select the most appropriate tools for their modeling. Recommendations for further improvements to the viability of these systems are also provided.

Design and realization of a descretized PV system with an improved MPPT control for a better exploitation of the PV energy

The performance of photovoltaic (PV) modules is affected by the increase in their operating temperature. Accurate and reliable prediction of the PV module operating temperature (T_M) is useful for evaluating their electrical performance and productivity and predicting the maximum power point under changing weather conditions. Based on experimental measurements collected during 1 year by a station placed in a semi-arid coastal region, in Morocco, several models are investigated to predict T_M of polycrystalline PV modules. Starting from existing models in the literature, models have been developed to improve the accuracy of their prediction in the studied climate. The results show that irradiance (G) and ambient temperature are the most influential parameters for estimating the T_M of the studied PV module. The accuracy of the prediction can be improved by introducing the effects of V_w and RH in the proposed models. In addition, using the physical parameter extraction results of the studied modules, equations are established to approximately predict their behavior as a function of T_M and G. Taking into account the proposed T_M models, these equations describing the physical parameter evolution can constitute an important basis for predicting PV module performance under real operating conditions.

In this paper, a converter DC/DC for power applications is developed: battery charger for photovoltaic system, vehicle charger, helicopter power supply. It consists of using a Full-Bridge DC/DC converter which is controlled by a new analog control, designed and analyzed during this work. This control has been dimensioned so that it ensures a smooth switching of the transistors, which makes it possible to reduce the losses of power, and consequently increases the efficiency of the overall system. It allows also to increase the current and to lower the voltage to the desired value; a large generated current which reaches 50A.

In this experimental study, we present a design and built of a single pass solar air heater with glazing and corrugated sheet metal absorber. It is an experimental prototype with a collector area of 1m ² , manufactured locally using simple materials with very affordable costs. Solar irradiation, inlet and outlet air temperature of the solar heater, as well as the average temperature at the absorber surface are measured. The effect of collector tilt angle on its performance is characterized. The average efficiency achieved is of about 41.2 % with air temperature difference of about 22°C, between the input and output of the solar heater.