Proceedings of the 4th International Conference on Electronic Engineering and Renewable Energy Systems—Volume 1

Our research aims to enhance the accuracy of Amazigh speech recognition through a data augmentation method. We employ Convolutional Neural Networks (CNNs) for speech recognition, utilizing Mel Spectrograms extracted from audio files. This study focuses on recognizing the first ten Amazigh digits. Our approach centers on refining recognition accuracy and augmenting data by modifying the number of bands in the filter bank. Additionally, our study involves 42 speakers who have participated, applying a speaker-independent approach throughout the experiments. We conducted three experiments using original data and three additional experiments using a combination of original and augmented data. The augmentation method involved changing the number of bands in the filter bank. Through these experiments with different CNN models, one model exhibited a 2.89% increase in accuracy, contributing to the overall improvement in speech recognition accuracy, with the highest achieved level reaching 95.88%.

Cognitive radio-Internet of Things is a new technology that has come to light in recent years. It refers to the integration of Cognitive Radio Networks capabilities within the Internet of Things devices. CR-IoT networks are designed to solve the spectrum scarcity problem in wireless communication systems. It is a powerful key solution that permits Internet of Things objects to overcome the challenges of the limited available bands of the radiofrequency spectrum; making its utilization more efficient and smoother and without harmful interference with other devices. CR-IoT networks enable dynamic spectrum access for Internet of Things devices which are defined as secondary users to opportunistically use the spectrum when it is not being used by primary users. This work stands as an open door to the world of Spectrum sensing in CR- IoT environment. It presents a bibliometric analysis that explores the landscape of research on spectrum sensing in the context of cognitive radio-based Internet of Things.

The Internet of Things (IoT) is undergoing rapid evolution, promising to connect billions of devices in the near future. This technological expansion presents numerous challenges, especially in maintaining diverse Quality of Service (QoS) standards across various applications. Our research introduces an innovative approach to evaluate the performance of the OneM2M platform’s middleware, crucial for the effective management of the expanding IoT landscape. We developed scenarios that simulate heavy load conditions on the platform, utilizing both uniform HTTP traffic and realistic IoT communication patterns. These scenarios aim to rigorously test the platform’s ability to cope with high network loads and genuine IoT operational conditions. Our approach is designed to provide a comprehensive assessment of the OneM2M platform's capability in handling different traffic types and volumes, an essential factor for its deployment in real-world IoT contexts. The results from these evaluations are critical in identifying the strengths and shortcomings of the OneM2M platform under various stress scenarios. These insights are valuable for enhancing the platform’s scalability, efficiency, and adaptability, ensuring that it can maintain robust QoS in the face of complex and dynamic IoT network demands.

Being propelled by the fourth industrial revolution (Industry 4.0), IoT devices and solutions are well adopted everywhere, ranging from home applications to industrial use, crossing through transportation, healthcare, energy, and so on. This wide use of IoT has not gone unnoticed, hackers are tracking the weakness of such a technology and threatening them continuously. Their security at various levels has become an important concern of professionals and researchers. This issue takes more risk, especially with the IoT variants, IIoT (Industrial IoT) and MIoT (Medical IoT). Many existing security solutions are adapted and proposed for addressing IoT security. In this paper, we are interested in exploring blockchain technology and we make a comparison of three free Blockchain platforms towards their applicability for MIoT context, namely Ethereum, Hyperledger Fabric and Corda. In general, Blockchain technology provides a decentralized, autonomous, trustless, and distributed environment. It is challenging to find a Blockchain platform that fits the MIoT context and performs well in terms of security. The retained platform should be deployed smartly to avoid its practical drawbacks related to energy-consuming and excessive computing.

This article presents a study that quantifies the effect of the human body on the measuring probe of a dosimeter, covering frequencies from 2G to 5G mobile networks. The study analyzed a human body model phantom using CST Microwave Studio, with a perfect plane wave as a far-field electromagnetic source. The electric field in the vicinity of the phantom was analyzed, considering frequency, polarization, and distance. The obtained results show the existence of a statistical rule that corrects the influence of the human body on incident EMF.

This study delves into the crucial issue of reducing energy consumption within green computing by examining the energy usage of well-known applications and comparing them based on their complexity. Utilizing PowerAPI, an advanced system-level library, we measure the energy consumption of these applications, with a particular focus on cryptographic hashing functions, notably the KECCAK function from SHA-3. Our preliminary findings clearly demonstrate how different algorithm implementations impact energy consumption. We introduce an innovative framework for energy monitoring, assessing popular applications and their energy usage while considering factors like CPU utilization. The aim is to identify energy inefficiencies in software. Our study not only investigates the energy aspects of hashing functions but also aims to promote the development of energy-efficient software for the future, aligning with the goals of green computing.

An effective planning and management of deadlines are crucial for global software development (GSD) project success. Establishing a schedule is essential, and one effective method is to propose a planning platform that uses a constraint satisfaction solver. This approach helps project managers analyze the feasibility of their projects and create effective schedules, considering all the variables of global software development. Our paper presents a systematic mapping study that provides a comprehensive summary of existing research on the aspects, techniques, and challenges related to planning and managing deadlines in GSD contexts, with a focus on studies that propose models using constraint satisfaction solvers. We found that successful deadline management requires thorough planning, including effective task allocation, management of requirement changes, risk management, and effort estimation. Consequently, our systematic mapping will concentrate on these four aspects, and we will examine whether there are articles that propose solutions using constraint satisfaction solvers.

In this paper, we utilize a previously published fixed point result to investigate the resolution properties of integral equations. Specifically, we explore the existence, uniqueness, and convergence of solutions using an iterative approach. Additionally, we provide an illustrative example to demonstrate the existence of an exact solution and compare it with the approximate solution obtained through the application of the main result from our previous publication. Finally, we analyze the differences between the exact and approximate solutions through a tabular representation of values and a corresponding graph.

Managing illnesses, treatment schedules, and hospital services represents an inherent and complex challenge, especially with sensitive treatments like morphine. This study presents a cross-sectional analysis aimed at identifying the needs of patients and healthcare professionals in order to develop innovative solutions that effectively meet these needs while ensuring high standards in terms of security and usability. For this, we have developed the HealthCare application, which stands out for its holistic approach to simplifying communication and coordination between patients, doctors, and hospital services. The app offers five key features: personalized user profiles for patients and doctors, appointment management for scheduling and tracking, medication reminders to improve compliance, and a specialized module for morphine management. HealthCare offers a user-friendly and secure interface, emphasizing data privacy and ease of use. By leveraging technological advancements such as artificial intelligence and the Internet of Things (IoT), the application strives to improve healthcare delivery. Supported by testimonials and case studies, our project also involved exhaustive benchmarking, comparing HealthCare to other similar solutions on the market. The results of our benchmarking demonstrated that HealthCare is positioned as a leading solution in terms of user friendliness, data security, features offered, performance, and reliability. This project aims to maximize benefits for patients and doctors, thereby improving the efficiency and reliability of hospital services. The HealthCare app represents a promising advancement in optimizing patient care and promoting collaboration within health systems.

Potato late blight is a devastating plant disease that significantly impacts agricultural productivity worldwide. This study evaluates the effectiveness of different data augmentation techniques in enhancing the performance of the YOLOv8m model for disease detection. The model was trained and tested using datasets augmented with brightness, contrast, Gaussian blur, and a final one that combined all of them. The results show that while individual augmentations to the original dataset slightly improved the model's performance, contrast manipulation substantially improved the model's precision, recall, and robustness, achieving a maximum mAP50 of 92.4% and a maximum F1 score of 86%. These results suggest that a well-tailored and optimized preprocessing pipeline greatly benefits the model's disease detection capabilities while reducing the computational load in incremental learning. This research sets the pathway for further developing and integrating these technologies to provide customized solutions for agricultural stakeholders.

Natural Language To Code Generation (NL2CG) represents a crucial intersection between human communication and programming, enabling the translation of human-readable instructions into executable code. This paper presents a systematic mapping study that explores the landscape of NL2CG techniques within a specific domain of application. The study investigates publications spanning the years 2013–2023, analyzing research trends, methodologies, and application domains. To keep the relevance of searched studies, my search is directed to studies published from 2013 to 2023. This systematic map contributes a comprehensive overview of the current state of NL2CG in a specific application context, shedding light on emerging trends, identifying gaps, and providing valuable insights for future research directions.

This comprehensive study assesses the capabilities of Whisper, an advanced open-source machine learning model, in transcribing Moroccan dialect from audio recordings. By employing both the base and large versions of the Whisper model, we analyze their transcription accuracy and their ability to understand the linguistic nuances of Moroccan dialect. Our findings contribute valuable insights into the strengths and limitations of these models, with implications for enhancing speech recognition technologies in diverse and dialect-rich linguistic settings.

In this paper, we study the performance of the protection relay IDMT (Inverse Definite Minimum Time) installed in a meshed electrical power grid in the presence of the statcom. This document presents the proposed models of the statcom and the IDMT relay used in this study, a comparison of the short-circuit current values and the tripping times values with/without the statcom for all faults type. The results are obtained by running simulations on MATLAB/Simulink.

This work aims to study the protection and reduction of perturbations that can affect the operation of a solar photovoltaic system. One potential disability is the behavior of a solar photovoltaic systems system connected to the LV or MV (low voltage, medium voltage) power grid in the face of transient regimes, in particular voltage dips or frequency variations. This study aims to simulate different dips and the system's instantaneous or delayed disconnection in the event of an error (voltage dip or low frequency). Validating and simulating the results obtained was done using MATLAB/SIMULINK.

This paper describes and investigates the performance of a new hybrid wind turbine system for water pumping. The proposed system uses several small wind turbines, instead of one turbine, to generate the electrical energy needed to power a 1.5 kW pump. Firstly, a techno-economic analysis of the systems showed that there are economic and technical advantages between one or more wind turbines to ensure the energy autonomy of a pumping system. Subsequently, the characteristics of the hybrid system were determined, and it is shown that the system can be optimized. A MPPT controller based on Perturb & Observe was tested and applied on the studied system to determine the appropriate MPPT algorithm. Simulation results verify the effectiveness of the proposed system and control algorithm.

Mastering the thermal behavior of an electric machine is a primordial step in its pre-sizing. Indeed, the optimization of the geometry of this one requires limiting the temperature in the active parts of the machine. This is why this article provides thermal modeling of the active part of the permanent magnet synchronous machine (PMSM). This machine is used for rim driving the turbine of an under-development pump. The machine is supposed to be cooled by water that flows through the submerged airgap, while natural air circulates on the external side. The geometry optimization algorithm considered the temperature in slots (which are the hottest area) as a constraint. A slot temperature of less than 155 °C was chosen, and the resulting geometry from this study perfectly fulfilled the specifications. The transient thermal behavior was studied and simulated for an optimized machine to highlight the sensitivity of the temperature of active components to different heat sources within the motor. It is shown that a.

This paper proposes the utilization of a Parallel Active filter (PAF) controlled by two different methods, proportional-integral (PI) and hysteresis controllers, in combination with L-filter, in order to enhance the power quality by eliminating the harmonic component generated by photovoltaic gen-erators in electrical grid. A PQ theory is used to extract the harmonic component of the injected current into the power grid. However, the quality of injected current is influenced by the solar irradiance levels. Hence, this study aims to investigate the influence of solar irradiance on the Total Harmonic Distortion of the injected current (THDi) and to conduct a comparative analysis of hysteresis and PI controllers in Grid-Connected PV systems.

The development of vehicle-to-grid technologies has been aided by the increased mobility of electric vehicles. Bidirectional power flow is made possible by vehicle-to-grid, or V2X, technology, which connects an electric vehicle’s battery to the other side. A specialized electric vehicle battery charger that enables bidirectional power transfer between the electric vehicle battery and the other side is necessary for the implementation of V2X technology. Typically, the main parts of an On Board charger are AC/DC with power factor correction (PFC) and DC/DC converter. In this paper, we focus on the second part which is the DC/DC converter, especially the bidirectional LLC resonant converter due to its high-performance efficiency. The resonant converter is controlled by varying the switching frequency; its formula can be extracted using the transfer function of the converter. A resonant DC/DC converter consists of 4 blocks the first one is the bridge composed of 4 switches its role is to generate a square wave signal, then a resonant tank (composed of Lm, Lr, and Cr), and after that, the signal is transferred to the second side via the transformer, which steps up or down the voltage, and then the switch bridge rectifier transforms AC voltage to DC allowing the bidirectional flow of power. This work aims to find the best compromise between resonant tank size and the control strategy of switching frequency.

This study introduces three controller types: field-oriented control with PI-controller, field-oriented control with Fuzzy-PI controller, and Backstepping controller. These controllers are implemented in the control system of a permanent magnet synchronous motor (PMSM) driven by an inverter designed for electric vehicles. The primary objective is to examine and compare the performance of the proposed PMSM drive with these controllers. The overall development strategies are outlined, and an illustrative example showcases the features of the control approach applied to a 0.2 kW PMSM drive. Finally, a comparative analysis is given to evaluate the effectiveness of the three distinct controllers, highlighting their performance in terms of reference tracking, ripple mitigation, and response times in both speed and torque profiles.

Today, investigation into renewable energies is endorsed by the rapid increase in electricity demand and the high cost of fossil fuels. The utilization of wind energy (WE) has seen significant growth. However, installing wind turbines for research and educational purposes presents challenges due to space constraints and maintenance requirements. Wind Turbine Emulators (WTEs) at the research laboratory level offer many advantages, including freedom from space constraints and the potential to develop improved control techniques independent of meteorological conditions. This study aims to design a WTE capable of accurately simulating the nonlinear behavior of real wind turbines, incorporating various turbine components such as blades, slow shafts, gearbox, and controller elements. The functionality and performance of the studied WTE are assessed under various wind speed conditions through numerical simulation.

This paper presents an overview of adaptive backstepping control for permanent magnet synchronous motors (PMSM). A combination of nonlinear control and adaptive mechanism is employed to ensure accurate tracking rotor speed, as well as direct current. Lyapunov-based design is used among this control strategy to enhance stability, even under uncertainties. The key advantage of adaptive backstepping mechanisms lies in their ability to adjust their parameters to compensate the variations in motor parameters, load fluctuations and external disturbances. The following study includes the state model of surface-mounted PMSM in the d-q frame, adaptive backstepping controller design to precisely estimate the stator resistance and magnetic flux on real-time, as well as the load torque, stability analysis. Several simulation scenarios in MATLAB/Simulink software tool are presented to validate the effectiveness of adaptive backstepping control, highlighting its potential as an effective control strategy for PMSM. The simulation results demonstrate the favorable dynamic performance of controller, providing accurate monitoring of rotor speed and currents, and disturbance rejections.

A wind turbine system with effective control can improve the controller and increase the amount of energy produced. As this work begins with a readjustment of some available control approaches, most analysis and research has taken place on the classical PI controller. Furthermore, the dynamic aspects of wind and aero-turbine are not taken into consideration, so their enhancement and performance are powerless. Advanced controllers are presented here in such a paper for enhancement; beyond that, the goal is to use wind turbine dynamics as a quantification tool. This paper presents the improved NSSFC and NDSFC controllers. The control target for the evaluated wind speed is to attain the optimum wind harvested energy while reducing mechanical loads. The MATLAB/SIMULINK environment was used to test and evaluate these techniques. The results also show the difference in control between the TSR and P&O methods, which will be the first step to open the space for working on the nonlinear techniques and the PI controller.

In this study, we present a simulation of a photovoltaic system aimed at efficiently regulating hydrogen production through water electrolysis, while adhering to electrical grid connection standards. This system relies on PWM controls, enabling us to maximize both energy and hydrogen production. The coupling of the photovoltaic system with the grid is achieved through a phaselocked loop (PLL), synchronizing the frequency and phase of the photovoltaic system’s inverter with those of the grid. This synchronization allows the photovoltaic system to inject energy into the grid, thereby preventing issues such as energy quality disturbances or damage to photovoltaic system equipment. Simulation results confirm that managing the energy produced by photovoltaic panels aims to achieve two main objectives: providing alternative energy via a DC/AC converter that meets the electrical grid’s needs and using excess photovoltaic energy to produce clean hydrogen.

This paper presents the equilibrium between accuracy and computational cost through artificial neural network (NN) in order to obtain models of nonlinear systems by Hammerstein Wiener (H-W) structure based technique of identification. The seek for optimal parameters is carried out by the back-propagation method using gradient descent. Emphasizing the enhance accuracy, a method to minimize the number of parameters of the H-W structure is provided. It is proved by minimizing the number of synaptic weights the result can be the same in the H-W structure. This method is substantiated in two leading assumptions: a) design (choose linear functions activation) and b) training (initialization of equal values block by block). Eventually, to decrease the number of parameters an algebraic operation is implemented.

This paper presents the architecture and the implementation of a hardware-in-the-loop (HIL) system designed for studying synchronous buck converters, also for the development and testing of control strategies. Starting with modeling and discretizing the synchronous buck converter, the work explains the process of constructing a cost-effective HIL system using the Texas Instruments F28069M LaunchPad™ development kit. The proposed HIL system enables real-time visualization of transient and steady-state responses with a LabVIEW Graphical user interface.

Recent advancements in energy technology have led to increased interest in DC microgrids as viable solutions for efficient energy management, particularly in scenarios involving renewable energy integration and distributed generation. However, the dynamic nature of renewable energy sources and the complexity of microgrid operation present significant challenges in achieving optimal performance and resource utilization. In this context, this paper investigates the application of fuzzy logic control as a promising approach to enhance energy management within DC microgrids. Fuzzy logic control offers advantages such as adaptability to nonlinear and uncertain systems, making it well-suited for addressing the variability and intermittency inherent in renewable energy sources. Furthermore, to ensure the effective design and operation of DC microgrids, theoretical methods are employed for system sizing. These methods take into account factors such as load profiles, renewable energy availability, and storage system characteristics to determine the optimal configuration and capacity of components within the microgrid. By integrating fuzzy logic control with theoretical sizing techniques, this study aims to overcome key operational challenges and contribute to the advancement of sustainable energy solutions. Through comprehensive analysis and simulation in MATLAB/Simulink, the proposed approach seeks to provide insights into the performance and effectiveness of fuzzy logic-based energy management systems in DC microgrid environments. This research makes a significant contribution by demonstrating the viability of a DC microgrid and developing an energy management system that will be implemented in a future scaled-down laboratory prototype.

This paper presents the control of a fully active Fuel Cell (FC)—Supercapacitor (SC) hybrid system using sliding mode control (SMC) theory. The suggested SMC technique is developed based on the system’s nonlinear model, utilizing Lyapunov stability design techniques. The proposed control technique enables tight regulation of the DC bus voltage, high tracking performance of the SC and FC current to their references and ensures asymptotic stability of the closed-loop system. Numerical simulations are provided to demonstrate that the proposed technique achieves all desired objectives.

The Phase-Shifted Full Bridge (PSFB) converter is a prominent choice for medium to high-power DC-DC applications, emphasizing efficient power conversion, high-power density, and isolation, particularly in applications like renewable energy technologies and electric vehicle charging systems. The PSFB can achieve excellent efficiencies, attributed to its soft-switching capabilities without the need for additional components. However, traditional PSFB converters encounter several challenges, specifically secondary side ringing, spikes, and duty cycle loss. These challenges can significantly impact the performance and reliability of PSFB converters. To solve these issues, researchers often employ techniques such as snubber circuits. This study investigates secondary side voltage problems and evaluates the effectiveness, efficiency, and complexity of C-only output snubber, RCD snubber, and Active snubber clamping circuit configurations using MATLAB Simulink models, providing insights for PSFB converter design and application in high-power contexts.

The present investigation carries out the study of requirements in the IEC and ANSI/IEEE international standards, analysis and recommendations of several specialists in transformer testing and impulse voltage testing systems for distribution and power submerged, in insulating liquid in order to carry out the sizing of an impulse voltage test generator for transformers up to 10 MVA, class 69 kV up to an altitude of 3,000 m above sea level (masl). The sizing indicates the step-by-step description of the characteristics and technical requirements that the system must have in order to carry out impulse voltage tests, both in peak voltage, as well as in the specific front and tail times. The results were verified through 87 simulations of 40,000 data each in Matlab/Simulink. Additionally, the simulations were evaluated with the results of high voltage laboratory tests on power transformers. The values obtained during simulation and laboratory tests are excellent since the deviations are less than 3% in voltage level, 8% and 12% in front and tail times, when the latter have a tolerance of 30% and 20%, respectively. This study provides the methodology and exemplifies the sizing of the main components of impulse voltage testing systems for distribution and power transformers.

The wide use of lithium batteries in electric vehicles highlights the necessity for precise state estimation and effective temperature control to ensure safety and reliability. Thus, developing accurate battery models is crucial for optimizing Battery Management Systems (BMS) and enhancing battery performance and longevity. This paper introduces an Electro-thermal Equivalent Circuit Model (ET-ECM) tailored for Lithium-Ion batteries, focusing on efficient estimation, observation, and diagnosis. Additionally, it compares the effectiveness of Linear-State Observer and Sliding-Mode Observer for estimating State of Charge (SOC), core, and surface temperatures.

The increasing share of renewable energy sources in the global energy mix highlights the challenges associated with maintaining grid stability and overall balance. This elevates the significance of creating suitable strategies, beyond what it would have been under other circumstances. This paper introduces the design, modeling and simulating of a micro-grid system consisting of 10 buses operating at medium voltage to leverage distributed generators, efficient consumption of energy, the robustness of the power grid, and the integration of innovative technologies. The objective of this initial study is to introduce the system and conduct load flow analysis for convergence and fault analysis to assure stability across various scenarios, namely high or low energy demand with and without distributed generators. The results show the robustness of the proposed system under parameter variation and external disturbances. Furthermore, the proposed system, initially comprising 10 buses, can be expanded by adding more buses while maintaining its original parameters.

This research explores the influence of current unbalance on technical losses in low-voltage distribution systems, introducing a novel and comprehensive approach to analysis and mitigation. Using CymDist software for detailed system modeling, the TOPAS 1000 grid analyzer for real-time monitoring of current variations, and MATLAB/SIMULINK to determine the functional relationship between the loss factor and the unbalance factor, our methodology provides an accurate and dynamic assessment of the impacts of unbalance. Through comprehensive data analysis and simulation studies, significant correlations between unbalance and energy losses are identified and analyzed. The study proposes practical measures for service providers to improve grid reliability and sustainability. By focusing on proactive management strategies and optimization techniques, this research makes a valuable contribution to the field of electrical engineering, offering concrete recommendations for improving system efficiency and reducing technical losses.

Hydrogen-powered electric vehicles (EVs) represent a significant advancement in clean transportation, utilizing innovative energy production and storage solutions. This paper proposes a robust control strategy for such systems, employing a combination of frequency splitting and the super-twisting algorithm (STA). The investigated system comprises a fuel cell (FC) integrated with a hybrid energy storage system (HESS) that utilizes both batteries and supercapacitors (SC). The DC-DC boost converter optimizes FC power output, while bidirectional DC-DC converters facilitate seamless power flow within the HESS for charging and discharging. This control strategy aims to overcome limitations inherent to classical PID control methods. Numerical simulations conducted in Matlab/Simulink validate the proposed approach's efficacy in mitigating chattering and over and undershoot phenomena under dynamic load conditions. The results conclusively demonstrate the effectiveness of the control strategy in enhancing overall system performance, enabling well-regulated power exchange, and achieving improved efficiency, reduced power losses, and enhanced system stability.

Hybrid Energy Storage Systems (HESS) have gained significant interest due to their ability to address limitations of single storage systems. This paper investigates the performance of two HESS topologies (Semi-Active, and Full Active) under a novel control technique based on the Super Twisting Algorithm (STA). The STA offers advantages over classical PI controllers in terms of improved response time and higher efficiency. Comprehensive simulations and analyses are performed using Matlab/Simulink environment to demonstrate these improvements. A comparative study has been further conducted to evaluate the configuration dynamics, including response time, and overall performance of the different HESS topologies.

This paper presents a comparative study of four control techniques: proportional-integral (PI) control, sliding mode control (SMC), Lyapunov function-based control and model predictive control (MPC). The proposed approach uses performance evaluations under dynamic conditions, including variable load profiles. Evaluation criteria include response time, reference tracking, stability and computational efficiency, implemented using Matlab/Simulink as the simulation environment. In addition, the paper proposes a comparison of these control techniques on the basis of their respective advantages and disadvantages, while identifying the most appropriate applications for each controller in the context of hybrid fuel cell/battery systems. Particular focus is placed on their suitability in the context of fuel cell electric vehicles, given their inherent dynamics.

Soil moisture is a critical variable in many fields of study, including meteorology, hydrology, and agricultural sciences. In this latter, surface soil moisture (SSM) is crucial for plant growth and development and consequently for yield estimation. Synthetic Aperture Radar (SAR) can be a reliable and trustworthy data source for SSM inversion through the use of empirical, semi-empirical and physically based models. Each of these methods has its own strengths and limitations, depending on the specific application and the environmental conditions. Likewise, there has been a recent surge in attention towards the use of machine learning regression algorithms in the SSM inversion process from SAR data. This work aims to assess the effectiveness of the two algorithms neural network (i.e. single-layer artificial neural network (ANN) and deep neural network (DNN)) for retrieving SSM by utilizing data gathered from diverse rainfall and irrigated (sprinkler) wheat fields located in Tunisia and Morocco. The comparison between predicted and measured SSM showed that the best retrieval results were obtained using sentinel-1 data at VV polarization with R of 0.75 and 0.76 for ANN and DNN, respectively. The RMSE was about $$ 0.05 \, {\text {m}}^{3}/{\text {m}}^{3}$$ 0.05 m 3 / m 3 for both algorithms. Overall, the performance of single-layer ANN mimics the highly complex multi-layers DNN in terms of statistical results at VV polarization.

The worldwide use of photovoltaic (PV) energy, a prominent renewable energy source, has significantly increased in recent years due to its numerous benefits. Agrivoltaic (AV) systems, which integrate agricultural and electricity production by placing solar modules several meters above the ground, are gaining popularity in renewable energy and farming sectors. The objective of this study is to assess the solar energy production that can be generated in the Berkane region in eastern Morocco, utilizing cultivated land for potato cultivation. For an installation capacity of 563 kwp the simulation by PVsyst software reveals a system production of 1.02 GWh/yr with a performance ratio of 0.863. This indicates that the grid-connected system has an effective capacity to convert incident solar energy into electricity. Furthermore, an economic analysis has shown that the average levelized cost of electricity production over the lifetime of an energy production installation amounts to 0.06 $/KWh.

This study proposes an artificial intelligence (AI) model that leverages B-scans for image classification, specifically targeting the accurate estimation of buried root organ orientation. The Convolutional Neural Network model exploits the intrinsic information within these images to analyze delicate features, enabling a rigorous classification of root orientations into 37 distinct categories for horizontal orientation ( $$\alpha $$ α ) and 19 categories for vertical inclination ( $$\beta $$ β ). This approach aims to significantly enhance the accuracy of root orientation estimation in subterranean environments, holding promising potential for practical applications, particularly in precision agriculture.

Water is a limiting factor for plant production and water scarcity has a significant effect on reducing plant growth, photosynthesis, and crop yield. During the last year, Moroccan growers suffered from the water scarcity caused by global warming. Area of vegetable production was reduced during the last years. Saving irrigation water is currently one of the main goals of the growers for the yield optimization in such conditions. The present work aims to evaluate the impacts of different irrigation regimes on photosynthesis (chlorophyll a fluorescence), stress index, plant growth and biomass and yield of lettuce (Lactuca sativa) grown under the semi-arid and Mediterranean conditions of the northeast of Morocco. The experiments were conducted in an open-field commercial farm of lettuce (var. Romain), located in Nador province. Three irrigation regimes were used as a treatment: (i): grower irrigation (Control); (ii) cutting 25% (Tr-75%) and (iii) cutting 50% (Tr-50%) of the grower irrigation amount. Split plot was used for the experimental design. A Digital soil moisture sensor was used in this experiment. Measurements of plant growth were measured each 2 weeks and the plant yield, fresh and dry weight of roots and leaves were recorded at the end of the experiment. Physiological parameters such as leaf relative water content (RWC) and leaf chlorophyll fluorescence were taken. Results showed significant differences for plant growth, yield, leaves, and roots fresh and dry weight. The chlorophyll fluorescence data showed different performance of plant stress under the three irrigation regimes.

Plant diseases are not just a local concern but pose significant threats to global food security and the sustainability of agriculture. They result in substantial yield losses, compromise crop quality, and impose daunting economic challenges on farmers worldwide. Traditional disease detection methods, reliant on manual visual inspection, are not only labor-intensive but also time-consuming. However, the advent of Deep Learning (DL) technology, offered solutions to these issues. DL algorithms, particularly Convolutional Neural Networks (CNN), have emerged as powerful tools capable of analyzing images of diseased plants, and accurately identifying pathogens or abnormalities. This study represents the potential of DL in agriculture. By evaluating some of the CNN architectures, AlexNet, VGG16, InceptionV3, and MobileNetV2, utilizing a dataset of 38 plant diseases and healthy classes. The results are compelling where AlexNet achieved the best Testing Accuracy (TA) of 94.55%. For, MobileNetV2, InceptionV3, and VGG16 achieving a TA of 92.92%, 90.72%, and 90.23%, respectively.

This paper details the conception of an electronic nose tailored for the precise detection of Volatile Organic Compounds (VOCs) emitted by cannabis and tobacco. Leveraging the Grove—Gas Sensor V2, equipped with four chemically sensitive sensors, our device offers accurate detection capabilities. Through rigorous experimentation, we have demonstrated its efficacy in identifying VOCs from cannabis resins and tobacco with exceptional accuracy and reliability. Employing advanced machine learning algorithms, we’ve developed an odor recognition model to interpret sensor data effectively. Notably, our focus on performance optimization includes integrating stabilization techniques to minimize external factor-induced variations, enhancing detection accuracy and reproducibility. This innovation represents a significant step forward in VOC characterization technology, promising diverse applications across industrial and healthcare sectors.

Accurate prediction of reference evapotranspiration (ET0) is essential for effective management across various sectors, particularly in agriculture and irrigation. This paper presents a comparative analysis of three prominent machine-learning techniques, Support Vector Machines (SVM), Random Forest (RF), and K-Nearest Neighbors (KNN), utilizing a comprehensive dataset of meteorological variables. The study focuses on the eastern region of Morocco, specifically Oujda, where water resource management is of critical importance. Through rigorous experimentation and evaluation, our findings reveal SVM’s superior accuracy (R2 = 0.9981) in ET0 prediction when compared to RF and KNN. This superior performance underscores SVM's potential for precise ET0 forecasting, facilitating informed decision-making in water resource management.

Olive cultivation has witnessed remarkable expansion globally, particularly in regions like the Mediterranean Basin and Morocco. However, olive plants encounter various threats from diseases and pests, such as bacterial blight, olive knot, peacock eye spot, and Aculus olearius infestations, posing risks to productivity and profitability. Traditional disease detection methods, including visual inspection and lab testing, are time-consuming and frequently prone to inaccuracies. In response, recent advancements in Artificial Intelligence (AI), particularly deep learning-based models like You Only Look Once (YOLO), offer promising avenues for early disease detection and management. This paper presents an advanced approach to detecting and identifying olive leaf diseases, focusing on peacock eye spot and Aculus olearius infestations, using the YOLOv8.1.21 object detection algorithm. Leveraging a dataset of 37,200 images sourced from in-field photography and public databases. The model categorizes olive leaves into healthy and infected categories, enabling early intervention and enhanced productivity in olive cultivation. This study highlights the importance of harnessing advanced AI in agriculture, particularly in olive cultivation. By achieving a remarkable mean Average Precision score of 99.5%, the developed detection method for olive leaf diseases holds promise for substantial improvements in plant productivity and profitability in the agricultural sector.

The application of Machine Learning (ML) algorithms has become widespread in solving diverse data classification challenges, extending to domains like agriculture, notably in tasks like plant disease detection. These classification tasks, especially in agriculture, often involve high-dimensional datasets, prompting researchers to seek efficient solutions. However, such problems pose significant computational challenges due to the vast number of variable combinations. Support Vector Machine (SVM), known for its efficacy with high-dimensional datasets and numerical features commonly found in plant disease detection, emerges as a promising solution. While SVM demonstrates competence with default settings, its performance can be notably enhanced through parameter optimization techniques like grid search and cross-validation (CV). This study aims the optimization of the main hyperparameters of SVM and the extraction of the best parameters, in this case, the optimal parameters lead to satisfaction, where, the average values of 79.17, 80.24, 79.17, and 79.57%, were obtained for Testing Accuracy (TA), Precision, Recall, and F1-Score respectively.

In response to mounting challenges posed by climate change, soil infertility, water scarcity, and agricultural pests and diseases, the adoption of innovative cultivation methods becomes imperative. Hydroponics emerges as a promising solution, offering a departure from traditional soil-based approaches by delivering essential nutrients directly to plant roots through nutrient-rich water solutions. This study explores the efficacy of hydroponic systems in addressing water management challenges while leveraging technology to enhance agricultural production. Through a comparative analysis between hydroponic and soil-based cultivation methods, key growth parameters including height, root length, fresh weight, root weight, and leaf number were evaluated for lettuce, basil, and spinach. Additionally, water consumption patterns were assessed to ascertain the water-saving potential of hydroponic farming. Results reveal significant improvements in plant growth and water efficiency in hydroponic systems compared to conventional soil cultivation, with water savings reaching up to 80%, highlighting its remarkable potential for effective water management.

Being largely utilized as alternative sources of energy, photovoltaic (PV) panels are being deployed in many locations. However, those panels are extensively being prone to faults. A careful detection and diagnosis of those faults becomes a must for solar energy engineers, practitioners and decision-makers in order to extract the maximum of energy from them. The integration of artificial intelligence (AI) in the domain of solar energy management has become a trend in the few previous years because of the high potential provided by machine learning (ML) and deep learning (DL) algorithms in handling issues related to PV systems such as control and fault detection and isolation (FDI). In this paper, TinyML models for defect classification of photovoltaic modules based on electroluminescence (EL) images were developed. An open-source platform, Edge Impulse, was used to design and deploy the models. Two convolutional neural network models (mobileNetV1 and mobileNetV2) have been used and compared. A dataset of EL with three anomalies was used to develop the models. The validation test shows an accuracy of 88.44 and 76.38% for respectively MobileNet-V1 and MobileNet-V2. The quantized (Int8) model has been integrated into a microcontroller (Arduino Portenta H7) for real-time application. The obtained results can be considered as promising and can be extended to other case studies with more defects and larger datasets.

This study investigates the Photovoltaic Thermal (PV/T) system’s efficiency in simultaneous electrical and thermal energy production, utilizing nanofluid coolants to cool temperature-sensitive PV cells. Filling a gap in existing models, a machine learning approach is examined to predict electrical and thermal power outputs specifically for a water-based PV/T system. Leveraging an extensive dataset of 15,540 water-based PV/T samples for training and testing, key thermophysical properties, time, and weather variables are considered as input features. The Random Forest Regressor model attains notable R2 values, reaching 0.9931 for electrical and 0.9852 for thermal predictions. Transitioning to nanofluid properties introduces a dynamic element, showcasing theoretical creativity and estimating electrical and thermal power outputs for Ag/water nanofluid-based PV/T systems. This study contributes to the advancement of predictive modeling in PV/T systems, demonstrating the model's accuracy and adaptability in the context of nanofluid-based energy production.

This paper investigates the use of multilayer neural networks (MLPs) to anticipate infections associated with totally implanted venous access systems (TIVAPS). Although these devices have revolutionized the treatment of patients requiring long-term injectable therapy, they also present a serious risk of infection. The study highlights the crucial importance of optimizing hyperparameters to improve the performance of the MLP model in predicting infections associated with totally implanted venous access port systems (TIVAPS). More specifically, the GridSearchCV classifier stands out for its balance between precision and recall, opening up promising prospects for the clinical management of TIVAPS complications. In addition, the use of the BaggingClassifier to optimize the model was explored, but its results were slightly inferior in terms of recall and F1 score compared with GridSearchCV. These results highlight the importance of further research to develop more effective prediction models, specifically adapted to the challenges of real clinical environments.

In efforts to reduce emissions, the production of green energy emerges as an effective solution. This study uses a wind-to-hydrogen system to evaluate the potential for green hydrogen production in Tan-Tan City, Morocco. The system is equipped with a 63 MW wind farm and six electrolyzer modules. The study has two objectives: firstly, to estimate the amount of hydrogen produced from January to December 2023 using this solution; secondly, to use the data generated by this system to train models and predict hydrogen production from January to April 2024. We employ two algorithms, Long Short-Term Memory (LSTM) and Recurrent Neural Network (RNN), for this purpose. The LSTM model demonstrated superior performance over the RNN model, as measured by the R-squared (R2) value, Mean Absolute Error (MAE), and Root Mean Square Error (RMSE) metrics. Additionally, we compared these two algorithms and concluded why the LSTM performs better than the RNN in our case study. This study and its results can provide a strong foundation for future research.

A theoretical framework is presented for studying the existence of topological states at the interface between two stubbed mesoscopic crystals (MCs). This study is based on the recently proposed novel mechanism based on band edge symmetry inversion around a flat band, i.e., when the width of the passband vanishes, while using only one stub per unit cell. The theoretical investigation of these states involves various approaches, including analyzing the topology of the bands based on the Zak phase and the symmetry of the band edge modes. Additionally, we consider the sign of the reflection phase between each MC and a waveguide to predict the existence and the position of such interface state. Subsequently, A straightforward deep learning network is proposed for inversely engineering such states with desired topological properties. This approach encodes the desired Zak phase of the bulk crystal in the sign of reflection phase, allowing the model to predict the corresponding geometry. Remarkably, our model achieves a 97.13% accuracy in performing such tasks on un-seen data.

To enhance the financial viability and optimize the design of solar energy projects, accurately assessing the performance and output of PV modules at the installation site is crucial. While various empirical and theoretical predictive models exist in the literature, their limitations and complexities often render machine-learning models a preferable alternative. This study evaluated the accuracy of three different machine learning models, Multiple Linear Regression (MLR), Artificial Neural Networks (ANN), and Extra Trees Regression (ETR), for predicting PV output, comparing their performance against a year’s worth of ground measurements from a CdTe PV module installed in a semi-arid climate location. The results indicate that all three models effectively simulated CdTe PV production based on meteorological parameters as inputs, achieving R-squared values exceeding 0.99 when comparing predicted outputs to actual measurements. Further analysis of the models’ performance has demonstrated that the ETR model outperformed the ANN, followed by the MLR, which tended to slightly underestimate the actual outputs.

The implementation of a multistate planning model for power generation-transmission expansion, as opposed to traditional independent planning frameworks, addresses the escalating energy demand and the need for efficient planning. This study applies a multistate DC-planning model to the 230 kV network of Ecuador’s National Interconnected System (NIS), aiming to minimize operational costs and the construction of new infrastructure. By employing a mixed-integer nonlinear programming (MINLP) approach and a neural network-generated demand forecasts, the electricity expansion spanning the period from 2023 to 2033 is analyzed. This methodology facilitates the delineation of expansion zones, thereby mitigating long-term costs.

In the realm of home energy management (HEM) system, the integration of demand response (DR) strategies with predictive modeling holds immense potential for optimizing consumption patterns. This article presents a comprehensive framework that combines reinforcement learning (RL) with predictive modeling using a Convolutional Neural Network—Long Short-Term Memory (CNN-LSTM) architecture. The CNN-LSTM model is employed to forecast photovoltaic (PV) power production and electricity prices, crucial factors influencing energy consumption decisions. The proposed approach considers various household appliance, including non-shiftable, power-shiftable, time-shiftable and Electric mobility devices, each represented as an autonomous agent, enabling decentralized decision-making tailored to specific operational constraints. By employing RL techniques, the model facilitates intelligent control, prioritizing the minimization of energy costs while simultaneously taking into account user comfort and appliance functionality. The simulation results validate the effectiveness and resilience of the proposed algorithm, showcasing a remarkable 36.7% reduction in electricity bills without compromising consumer satisfaction.

Accurate energy demand forecasting in the building sector is essential for sustainable energy planning and policy development. This paper proposes a Genetic Algorithm GA based methodology for forecasting energy demand in Morocco’s building sector until 2050. The methodology takes into account historical energy consumption data, socioeconomic factors, and urbanization patterns to represent the complex processes that drive energy demand. The model is evaluated against real International Energy Agency data, indicating a strong correlation coefficient (R2 = 0.9634) between projected and observed energy demand. Once validated, the model uses predicted socioeconomic data to assess habitable areas and then anticipates future power usage for buildings. The results show that energy demand increased progressively during the simulation period, with significant results projected for 2030, 2040, and 2050. By 2030, energy demand is predicted to increase by 84.33% over 2017, due to growing populations and economic development. Energy demand is expected to increase by 74.27% and 79.50% in 2040 and 2050, respectively, reflecting the continued growth of energy needs. These findings highlight the significance of strategic energy planning and sustainable infrastructure development in Morocco’s building industry, which promotes resilience, efficiency, and environmental sustainability.

Morocco, along with numerous other nations, is exploring the potential of electric vehicles (EVs) to reduce dependence on fossil fuels and mitigate the impacts of cli-mate change. This study undertakes the prediction of energy demand in the transport sector by 2050 utilizing artificial intelligence algorithms. Various scenarios regarding the integration of electric vehicles and their impacts on energy demand and CO2 emissions are examined. Initially, neural networks (ANN) are employed to generate efficiency indicators (EI), such as average vehicle kilometers traveled (AVKM) and stock vehicles (SV), from socio-economic indicators (SEI). Subsequently, the energy demand of the Moroccan transport sector from 1990 to 2017 is modeled, and this model is validated using real data. Subsequently, energy demand from 2017 to 2050 is predicted, analyzing six scenarios: the first scenario serves as the reference scenario with-out EVs. Following this, four scenarios entail the integration of EVs with increasing penetration rates (10%, 30%, 80%, and 100% of new cars being EVs each year). Finally, scenario six corresponds to the conversion of the entire car fleet to EVs.

Water, as a vital element, occupies a pivotal role in sustaining life, essential for the survival of all organisms. The challenges concerning the safety and accessibility of drinking water present urgent global concerns. This article examines the increasing impact of machine learning and the IoT on managing water quality. We explore IoT technologies that are used to collect water quality data in real time, as well as machine learning methods that facilitate effective data analysis. We emphasize the importance of these technologies for environmental preservation and public health by highlighting their useful applications in managing and monitoring water quality. These developments are expected to improve water management systems’ responsiveness, reduce contamination concerns, and encourage the sustainable use of water resources. Future water shortages and pollution will be major concerns, thus it will be critical to use IoT and machine learning to their full capacity in order to protect this priceless resource for future generations.