ABSTRACT. Los autocodificadores resultan de la combinación de dos redes neuronales: en codificador y el decodificador. El codificador tiene como objeto tomar los datos, generalmente, en dimensión muy alta y aplicarles una transformación no lineal para llevarlos a un espacio latente de menor dimensión. Luego el decodificador toma los datos codificados en el espacio latente y mediante otra transformación no lineal trata de recuperar los datos originales.  La denominación “variacional” ocurre cuando se imponen penalizaciones o restricciones que regularizan el espacio latente.  Por ejemplo, que el vector latente tenga una distribución normal con media cero y varianza uno. Esta restricción permite poder generar datos sintéticos a partir de muestrear la distribución simple de las variables latentes. En esta charla revisaremos variantes de VAE que imponen otras restricciones los vectores del espacio latente, como su cuantización, con el propósito de mejorar la generación de datos: que luzcan más realistas. Los VAE con vectores cuantizados son la base de los modelos modernos de generación de imágenes como Difusión Estable.

ABSTRACT. An important stage in recognizing Land Cover (LC) on satellite images is the determination of a set of features that can best describe it and help infer between types of covers. This article analyzes and selects texture features of the Gray Level Co-occurrence Matrix (GLCM) and Local Binary Pattern (LBP). The selection of texture features makes it possible to reduce the dimensionality to a relevant subset that provides information in improves the identification between types of LC. This in order to improve performance in semantic segmentation methods in satellite images. The images used for the analysis experiments corresponded to multispectral satellite images from the Sentinel 2 satellite of COPERNICUS of the European Space Agency (ESA). GLCM and LBP features are extracted from patches of different LCs of interest (agricultural vegetation, vegetation, soil, water, and urban). In addition, coefficients of variation of values of the features are obtained, and an analysis of the variation behavior between the different interest coverage and within the same type of LCs is carried out. Finally, relevant features were selected considering the behavior of the coefficient of variation and those obtained with the Relieff method.

ABSTRACT. The tequila and agave culture are important symbols of the national identity in Mexico. In the last decade, the production of tequila has increased and improved the care of Tequilana Webber Blue Agave. In this paper, we introduce a methodology based on the use of Unmanned Aerial Vehicles (UAV) as drones, image processing techniques and machine learning approaches for the detection of agave plants. Several landscape images of agave crops are collected from the Romita region in Guanajuato state. These images are processed to obtain orthomosaics which allow us to study all the target region of the agave crops. The orthomosaics are segmented using superpixels techniques in order to define features which allow us to recognise the agave plants. For experimental purposes we compare five segmentation methods in order to evaluate the performance for the detection of agave plants. Random Forest Algorithm is implemented for classification purposes. Our proposal achieves 97% of accuracy to detect the agave plants.

ABSTRACT. The detection of states in the driver is important for the prevention of traffic accidents, because, unfortunately, the accidents caused by undesirable state in drivers happen around the world and these must be dealt immediately. Drowsiness in drivers is one of the most important factors causing traffic accidents, especially on highways. In this paper, we propose models based on machine learning techniques to predict driver states by extracting ten facial features. We use the MediaPipe Face Mesh for the detection of relevant points in the face, and then extract ten facial features related to the driver’s state. We apply the machine learning techniques to effectively predict the four driver’s states, which are normal, talking, yawning, and sleeping. The machine learning techniques used are K Nearest Neighbours (KNN), Support Vector Machine (SVM) and Long-Short Term Memory (LSTM). The effectiveness of each technique is compared using two public databases.

ABSTRACT. Due to the increasing demands for efficient and flexible manufacturing processes in the era of Industry 4.0, the integration of augmented reality (AR) technology has emerged as a promising solution to enhance the training of assembly line operations. This article presents the design and implementation of an AR-based assembly line for Industry 4.0 applications. Utilizing Unity as the development tool, we created an AR application for training manufacturing engineering students and workers that allows them to visualize and interact with virtual components overlaid in the real-world environment using their mobile phones or tablets. Through a user-centered design approach, we optimized the user interface and interaction design to ensure an intuitive and user-friendly experience for workers and students. Furthermore, we discuss the implications that our findings have in the context of Industry 4.0 applications, highlighting the potential of AR technology in revolutionizing assembly line operations, thus contributing to the advancement of Industry 4.0 applications by harnessing the capabilities of AR technology to enhance assembly line operations, improving worker's productivity, students' training, and drive innovation in the manufacturing industry. The developed AR assembly line might offer a range of benefits, including improved efficiency, reduced errors, and enhanced training capabilities. The application has strong potential to be used to provide real-time guidance and performance feedback to facilitate seamless assembly processes.

ABSTRACT. Most biomedical research has focused on the study of ECG signals to diagnose heart or chronic diseases such as diabetes and hypertension, among others. Most heart-related diseases are identified by changes in the morphology of the waves and complexes that form the ECG signals, however, automated techniques for detecting ECG signal features result in expensive equipment. In this article, the authors present automatic feature detection and extraction techniques in real ECG signals using low-cost platforms and algorithms based on the discrete wavelet transform. The platform used to acquire ECG signals is the ADS1293EVM acquisition card and the MATLAB software is used to compute these signals. The developed program is evaluated with 30 ECG recordings of 6 seconds, having results in the detection of characteristics of interest greater than 90%.

ABSTRACT. Melanoma is an aggressive skin cancer that can rapidly spread to other parts of the body if not diagnosed and treated promptly. Current diagnostic methods include visual evaluation, biopsy, and histopathological analysis, but can be subjective and require significant time and resources. This work proposes the development of a melanoma classification protocol based on small and large MobileNetV3 architectures combined with two fine-tunning schemes. MobileNetV3-based models were used with transfer learning, and fine-tuning schemes were incorporated into the base model for melanoma classification. The best performance was achieved by the large MobileNetv3 architecture with the fine-tuning 2 schema. Training evaluation on 2003 images reported a successful mean of the area under the receiver characteristic operating curve score of 0.906. Additionally, the test on 223 images provided a reasonable score of 0.917. Both results were obtained using a stratified ten-fold cross-validation mechanism. The best model was implemented on two mobile emulators to analyze its feasibility according to power consumption, and the obtained mean of 0.45 mAh per image constituted a high-quality performance. Furthermore, it was implemented in a web app, and the average response time of 115.44 ms with an average of 15kb transferred over the network per image demonstrated efficient utilization of computational resources. Therefore, developing and deploying successful deep CNN models with transfer learning into limited-resource devices is possible as a helpful second opinion tool for early patient self-diagnosis of melanoma.

ABSTRACT. Abstract—According to the World Health Organization, breast cancer is defined as the abnormal and disorganized growth of cells in breast tissue. It is currently one of the biggest challenges facing health systems worldwide. Making a precise diagnosis is essential to offer the most appropriate treatment adapted to each patient condition. Computer vision-based tools could help the specialist in the diagnosis task. One major challenge in digital histopathology is the wide variation in staining tissue. This work assesses the performance of color normalization under different approaches, measuring its impact on the structures of interest related to the graduation of tissue samples. It is also analyzed how color normalization could improve relevant structures segmentation, particularly nuclei, an essential element in the Nottingham Graduation System followed by pathologists. Four methods of color normalization were implemented, and their respective structural and color-associated metrics (SSIM, PSNR, and colorfulness) were calculated. After color normalization, a segmentation process was performed using the STARDIST tool. In general, an improved segmentation was observed for the Reinhard and Stain-Net methods of color normalization.

ABSTRACT. The characterization of conductive and insulator materials plays an essential role in biomedical engineering. A technique is to measure the difference in conductivity of the domain and compare it with a reference material. The present work analyzes the resistivity difference between a copper and PVC tube. The data acquisition was obtained by the LatePanda device integrated into an Arduino board coupled to a phantom of 16 electrodes located on the periphery. Electrical Impedance Tomography (EIT) was used to measure the resistivity changes. The images showed a significant difference between conductive and insulating material. The obtained information from the test will be used in future works to determine the varying of resistivity amongst a healthy and one cancer tissue

ABSTRACT. This paper presents the most important characteristics and dimensional criteria when specifying a Battery Energy Storage System (BESS). Rated energy and power capacity values and their meaning in different measurement points are discussed. Both system and individual subsystem efficiency in different operation points is considered. Battery lifetime definitions are presented and their relationship to the above characteristics discussed. Finally, an example design process with a specification is presented.

ABSTRACT. This paper analyses the environmental impact of power generation plants in Mexico throughout their life cycle assessment (LCA) from the cradle to the grave. The studies were carried out in the SimaPro Software and using a functional unit of 1 kWh. The obtained results let to know which of the generation technologies used in Mexico is more polluting and at what stage of its life cycle, such as assembly, use, or disassembly, it pollutes more. With the obtained results, it is possible to know the optimal technologies to be installed and propose solutions to reduce the impact of the most polluting technologies. A global ranking with generation technologies was calculated. This ranking shows that the least polluting technology is nuclear power, followed by photovoltaic plants, while the most polluting technology is combined cycle plants.

ABSTRACT. Increasing penetration of behind-the-meter (BTM) resources in distribution systems is a prominent factor for inducing the death spiral in retail markets. Owing to realtime deviations in BTM generation, the energy procured by retailers in the day-ahead markets may no longer be sufficient. Thus, the retailer must procure excess energy at spot prices to compensate these deviations, which in turn increases energy costs for the customers, driving more BTM resource adoption and reinforcing the death spiral. In this context, this paper develops a framework for the electric retailer to procure energy-storage-as-a-service (ESaaS) to mitigate the operational uncertainty. Volume firming contracts are established between the retailer and utility-scale energy storage operators to minimize the retailer’s energy procurement costs in spot markets, thereby limiting energy costs for the customers in spite of high uncertainty introduced by BTM resources. This paper develops the mathematical framework for the volume firming contracts, obtains conditions for optimality and profitability of the contracts, and discusses implications on social welfare. Simulation results verify the effectiveness of the developed framework in improving the financial situation of the retailers and ESaaS providers under uncertain generation conditions of residential BTM resources.

ABSTRACT. A statistical analysis method is developed for estimating patterns of behavior and their amplification from the observed system response to random variations. Spectral proper orthogonal decomposition (SPOD) is used to examine gains and phase relationships between signals from simulated data. Based on frequency domain and Fourier analysis, the SPOD is employed to decompose forced oscillations into a set of space-time orthogonal modes. Such an approach improves the ability of Fourier analysis to study stochastically forced oscillations in complex power system models. Analytical criteria to describe the energy relationships in the observed oscillations are derived, and a physical interpretation of the stochastic modes is suggested. The methodology is illustrated on simulated random oscillations of a realistic test power system.

ABSTRACT. Fault detection and condition monitoring of induction motors (IMs) are of paramount importance for the industry. In this regard, researchers around the world have developed different methods to monitor and detect various types of damages in IMs, such as damaged bearings, misalignment, and broken rotor bars (BRBs), among others. In particular, the detection of a BRB has received special attention as it is difficult to detect at an early stage and can quickly evolve into catastrophic damages if not detected in a timely manner. As a contribution to this issue, this work explores the potential of the Hjorth parameters as indicators of BRBs using the current signals of an IM under different operating conditions, i.e., healthy (HLT), half BRB, one BRB, and two BRBs. Obtained results show that the Hjorth parameters are sensitive to the previously mentioned conditions, allowing the proposal of pattern recognition schemes for automatic classification. For this task, the k-means clustering method is proposed in this work because of its easy implementation. The obtained results demonstrate that the proposed method is reliable to monitor the IMs condition, reaching an accuracy of 98.75%

ABSTRACT. In order to facilitate the design of classical controllers, modeling functions for unknown control systems is an important goal in control engineering. Therefore, the Recursive Least Squares (RLS) technique is the most recognized method for system identification. In the presented work, the digital architecture of the reprogrammable RLS algorithm for the identification of control systems is proposed, reducing its components by developing the relevant equations for the calculation of the coefficients and minimizing the consumption of resources within a reprogrammable FPGA EP2C20F484C7N board. The proposed embedded system consumes 19% of the logic units and 29% of the multipliers of this device and can also work at a maximum operating frequency of 36.69 MHz.

ABSTRACT. The work describes the development of a three-level converter based on a switched source capable of generating typical fault signals present in the power grid. The process involves designing, simulating, and implementing the converter using an STM32 NUCLEO-F746ZG DSP board and an H-MOSFET bridge card acting as an inverter. The control circuit is implemented through Simulink, emulating various typical faults found in power grids. A VHDL-based data acquisition system using a GENESYS2 FPGA has also been developed to analyze and record the emulated fault signals. The generated faults include sag, swell, flicker, harmonics, and variations according to the IEEE 1159-2009 standard. The obtained results enable the evaluation of the impact of these signals on the converter and the electrical system, which is crucial for enhancing the quality and stability of the power supply and designing more reliable systems.

ABSTRACT. Short circuits in windings is a major factor contributing to the damage observed in electrical transformers, therefore, early detection during the initial stages is of vital importance to prevent more extensive damage. This paper proposes an approach for detecting short circuits through vibration analysis. The proposed methodology enables the analysis of various conditions, ranging from a healthy state to six levels of short-circuit turns in an unloaded transformer. To accomplish this detection, it is employed a combination of frequency domain features extraction after the application of the FFT, principal component analysis, and a classification technique such as K-nearest neighbors or Support Vector Machine. The proposed approach accurately determines the extent of damage present in the windings. The results show the effectiveness of this proposal in precisely identifying the severity of damage in the transformer.

ABSTRACT. Sea lions (Zalophus Wollebaeki) are a protected species, and effective monitoring is crucial for habitat preservation and behavioral studies. However, manual sea lion counting is laborious and error-prone. In this paper, we explore the use of two standard convolutional neural network models (YOLO-NAS and YOLOv8) for sea lion detection as a preliminary step toward automating the counting process. For this purpose, a dataset of images and videos of sea lions was collected in their natural environment in the Galapagos Islands. The results demonstrate that both models exhibit promising detection capabilities, successfully identifying almost all sea lions in the images. In particular, YOLOv8 shows to be more reliable in the detection of sea lions under challenging and complex conditions, while YOLO-NAS excels in the identification of a larger number of individuals, including those of a smaller size. These findings pave the way for future developments in automated sea lion counting tools, streamlining conservation efforts and advancing our understanding of this protected species.

ABSTRACT. In this work, a comparative evaluation of embedded devices for Natural Language Processing (NLP) in biomedical applications in Spanish is carried out, specifically in text-to-speech and speech-to-text algorithms. Several embedded devices have been selected to achieve this, including Jetson Nano, Raspberry Pi (4B, 400, and 3B+), and Latte Panda. This analysis focuses on key aspects such as execution time, CPU usage percentage, and RAM usage percentage. These criteria will allow us to compare the evaluated devices' performance and determine the most suitable NLP in biomedical applications in Spanish. This benchmarking is expected to provide valuable information to facilitate the selection of devices in future projects and applications within the NLP. This work will allow professionals and developers to make informed decisions and optimize their resources when implementing PLN solutions in the biomedical field in Spanish.

ABSTRACT. Ladybird beetles, also known as ladybugs, are a diverse family of small, brightly colored beetles with thousands of species around the world. Their geographic distribution and their impacts as invasive species on other endemic populations are still not well understood. The ability to accurately identify and study ladybird beetles is essential for effective management and conservation efforts. In this paper, we propose a novel method for the detection and segmentation of ladybird beetles in natural environments. Our approach combines the YOLOv8 object detection model and the "Snakes" active contour model to achieve precise detection and segmentation of ladybird beetles. We evaluated our method on a dataset of 2300 ladybird beetle images from the iNaturalist project, obtaining a DICE score of 84.82% and an IoU score of 73.73%. These results demonstrate the effectiveness of our approach in accurately identifying and segmenting ladybird beetles. Our method has potential applications in biodiversity research, invasive species management, and ecological monitoring.

ABSTRACT. The protein side chain packing is a well-known NP-hard computational problem and of prime relevance to the study of protein structure-function relationship and protein design. Central to the solution of this problem is the design of rotamer libraries. These libraries have improved with respect to the coverage of torsion angles, at the cost of becoming larger and more complex. However, larger libraries imply a larger size of the algorithm’s search space. To address this issue, we propose a method for designing rotamer libraries which considers pairs of consecutive amino acids, the resulting library of pairs leads to an exponential reduction of the size of the search space the algorithm needs to analyze. The proposed approach is based on the k-modes clustering algorithm to generate the library. The proposed method starts with a set of target protein structures and a desired coverage level of dihedral angles in a target dataset, then for each pair of consecutive residues in the target set, generates a set of pairs of rotamers that attains the pre-specified coverage level. We show that our proposed method generates good achievable accuracy when built on a set of 149 structures and tested over two sets of 65 and 373 protein structures.

ABSTRACT. This document presents the simulation results of a novel inverse kinematic-based controller which is time-invariant. The proposed controller addresses the person-following robot problem, in order to achieve a social robot motion a modulation of controller gains is proposed, to modify the velocity response of a differential drive mobile. According to the results, our proposal has good performance.

ABSTRACT. The rapid expansion of the Internet of Things (IoT), now with billions of mainly wireless interconnected devices, has brought concerns and challenges regarding the security of IoT networks and devices, as they are often vulnerable to attacks. In this article, a measurement campaign is carried out to investigate the path loss experienced by a portable IoT hacking tool, known commercially as Flipper Zero, when used in transmitter mode, to shed light on the maximum achievable distance at which an IoT device can still receive an eavesdropper’s signal above a minimum power level. The path loss measurements are performed in three different outdoor environments. The results show that the hacking tool transmitted signal can reach up to 15 meters with power above -90 dBm, which is still in the sensitivity range of many IoT devices, thus revealing potential vulnerabilities and security risks in real-world scenarios.

ABSTRACT. Intensity and color of the Light are essential factors for the proper growth of a crop. Currently, artificial lighting is used to grow indoors where there is no access to sunlight. In this work, the implementation of a novel and simple automated lighting system for an indoor growing prototype is presented: This system provides independently blue and red LED lighting with a light intensity of 130 PPFD for the growth of Lettuce "Baby Sucrine" variety. The prototype uses a PSO anti-windup PID controller to ensure that the light intensity remains constant. The color signal references to be applied into the crop are given in base of agricultural research in this topic. The light intensity is monitored through a web interface as well as the color reference. The interface displays lighting data for the last 24 hours. Results obtained from lettuce growing in the prototype are compared with a control crop grown under 130 PPFD white LED lighting with the same photoperiods.

ABSTRACT. The use of DC/DC converters with high gain has become indispensable in applications such as renewable energy, electric vehicles, DC microgrids and energy storage. Some requirements are high conversion ratios, resulting in high order converters. That is, converters with a greater number of reactive elements (inductors and capacitors) as well as switches. One of the challenges of these systems is the controller design, which must be able to regulate the output, with fast dynamics, mitigate the effects of parametric uncertainties and rejection to disturbances in the system inputs. The controller design requiere a controllability analysis to verify whether is state variable is adequate. This paper presents the controllability analysis based on switched linear systems of a quadratic Buck converter with reduced redundant power processing ($R^2P^2$) with an input LC filter. In addition, a comparison of energy processing based on graph theory against cascading structure is presented.

ABSTRACT. Originally, Lyapunov’s Theory was specifically conceived to analyze closed-loop systems with no control inputs. If systems with control inputs are considered, the approach based on Control Lyapunov Functions (CLF) can provide analysis facilities. A CLF is a candidate to be a Lyapunov Function (LF), thanks to the fact that its derivative can become negative through the appropriate choice of control input. The existence of a CLF is sufficient (and necessary) to asymptotically stabilize a nominal system. This paper presents a procedure through which a CLF can be obtained that allows proposing control schemes for voltage regulation in DC-DC Boost Converter modeled by Hamilton’s equations as a case study. Obtaining this type of regulators makes it possible to ensure the asymptotic stability of the system.

ABSTRACT. Fossil fuels are widely used for power generation, but they are increasingly scarce, polluting, and expensive. For this reason, the incorporation of energy generation from renewable sources is increasing at an accelerated rate in recent years in the electricity sector. The injection of large amounts of wind and photovoltaic energy into the electrical system must consider the changes that the stability of the system may undergo, being the frequency stability directly affected by the conservation caused by renewable generation and therefore strategies must be implemented to guarantee the safety and reliability in the operation of the electrical system. The main contributions of this paper are: 1) the implementation of an observer-based state feedback control, to achieve a convenient relocation of the power system poles and improve the frequency behavior of the system, so that it reaches the steady state without the presence of oscillations and in a shorter period of time; 2) the implementation of the super twisting controller to compensate for the intermittency of renewable generation in the power system and whose action contributes significantly to the system frequency stability, achieving rapid convergence in finite time and without steady state error. For the case studies, the results were satisfactory, demonstrating that both control techniques are highly efficient.

ABSTRACT. This paper is devoted to study a new statistical description based on the polynomial distribution function for the reliability of electric power systems. An electrical system can be represented as a structure composed of many components, parts and blocks, each of which ages and tends to fail. That is why preventing failures is especially important in electric power systems, where both industry and the population depend on the generation and constant delivery of energy. Any failure in the system affects a large number of consumers. Each system component is characterized by its time-dependent hazard function (or failure rate). Some of its statistical properties are discussed, method of moments and Weibull distribution are used for estimating the parameters. A simulation study is performed to compare the performance of each method of estimation. Finally, for the description of the reliability of electrical systems, it was found that the representation of failure times by a distribution function with polynomial failure rate is more suitable according to the Kolmogorov-Smirnov (K-S) criterion.

ABSTRACT. The integration of variable renewable energy sources (VRES) into power systems has brought about changes in the operational control requirements of electric power grid. Voltage regulation and management of reactive power resources are crucial to maintain power system efficiency, quality, reliability, continuity, safety, and sustainability. In this context, the power system operator is responsible for defining and coordinating the necessary actions to maintain both the voltage within the established ranges and an adequate reserve of reactive power to support contingencies and/or failures in the network. This paper evaluates the impact of defining the prioritized reactive power control mode for reliability on VRES. Two case studies are presented to compare the performance of voltage regulation with the three possible modes that are reactive power setpoint, voltage control, and power factor setpoint. The results show that prioritizing the reactive power control mode can significantly enhance the performance and reliability of power systems with VRES integration, particularly during dynamic and fault conditions. The findings of this study can provide useful recommendations and operational strategies for power system operators to improve the reliability and safety of power systems with VRES integration.

ABSTRACT. This article presents a didactic integration of the PowerLogic™ module into the laboratory at Universidad Polit´ecnica Salesiana (UPS). Research and analysis have been conducted to incorporate didactic methods that allow students to acquire academic knowledge and gain familiarity with the work environment, generating academic and industrial benefits. Additionally, equipment has been integrated, and research has been carried out on the IEC 61850 System Configurator® software communication protocol and Modbus TCP/IP.

ABSTRACT. The transmission grid transports the electricity that is traded between different suppliers and demanders in a market environment. Currently, these electricity suppliers have increased their generation investments in renewable technologies such as wind and photovoltaic, whose ideal locations are in places far from the major centres of demand, in this scenario is that it must have adequate transmission capacity, Otherwise, there will be an increase in the cost of electricity, or the collapse of the system, which is why it is necessary to have a plan for the expansion of the transmission network capable of finding the best configuration of the electricity network with the lowest investment cost to meet the load forecast and take advantage of the development of new technologies. In this paper, a model is proposed to evaluate the integration of battery-based storage systems in transmission grid expansion planning. The model is validated on the six-node Garver system and tested on the 24-node IEEE-RTS system.

ABSTRACT. Currently, electric power systems have been integrating massive amounts of renewable energy sources based on power electronics converters. This tendency has provoked new stability issues because of the interactions of controls in power electronics with passive elements in the power system. As a result, new oscillation issues have been reported, such as resonance problems in wind turbine generators (WTGs) with series compensation. This paper evaluates resonance problems in a WTG Type 4 (WTG4) using the impedance-based stability criterion due to phase-lock loop (PLL) control interactions. The sensitivity of the WTG4 to the bandwidth of the PLL is assessed using small-signal models and nonlinear time-domain simulations in Matlab/Simulink are conducted to support the results.

ABSTRACT. This paper presents a passivity-based control of a non-inverting buck-boost converter coupled to a proton exchange membrane fuel cell. The controller primary objective is to maintain a stable load voltage despite fuel cell voltage variations by assuring precise current control. To accomplish this, two feedback loops are used: an outer loop that generates the current reference based on the output voltage using a proportional-integral (PI) action, and an inner loop that tracks the current using the passive characteristics of the system. To improve the reliability of the inner current loop, an immersion-invariant (I\&I) load estimator is developed. The overall efficacy of the controller is determined by numerical simulations that demonstrate its capacity to manage load variations and maintain a stable, regulated DC output voltage despite voltage changes from the fuel-cell stack.

ABSTRACT. This paper presents two Neural Networks (NN) that estimate the wind speed and optimal mechanical speed. These use the Tip Speed Ratio (TSR) technique for Maximum Power Point Tracking (MPPT) in a simulated Wind Energy Conversion System (WECS) connected to the grid. The purpose of the NNs estimators is to provide the appropriate mechanical speed reference on the speed control loop despite the random behavior of the wind, avoiding the use of an anemometer from the cut-in speed. This is achieved by estimating the wind speed from the power of the wind turbine (WT) and mechanical speed of the permanent magnet synchronous machine (PMSM). The performance of the control applied to the converters is evaluated through a simulation in MATLAB/Simulink. The stability of the machine-side converter (MSC) and the grid-side converter (GSC) systems is demonstrated by applying a stepped wind speed profile and a random wind speed profile.

ABSTRACT. The implementation of renewable energies in electric power systems requires the integration of direct currentbased transmission technologies. In HVDC systems, half bridge modular multilevel converters (HB-MMC) are frequently used, however, a major challenge for DC-based power conversion and transmission systemsistheir protection againstshort-circuit faults on the DC side. Short-circuit fault currents have fast rise rates as well as high current amplitudes. Although the full-bridge topology (FB-MMC) is fault tolerant, it suffers from high switching losses. In this paper, the implementation of a hybrid HB/FB MMC with one path for the steady-state current and another path for the fault current is proposed. The fault current evaluation through the hybrid MMC is implemented in PSCAD/EMTDC.

ABSTRACT. Numerous applications use power conversion by applying DC/DC converters with wide transformation ranges and capacities to provide high current or voltage levels. Low voltage applications have been standardized to 48/24/12/5 V for various Electric Vehicle applications such as lighting, GPS, audio and air conditioning. However, for fast charging of electric vehicles, high voltage levels standardized to 380 V are required. Some requirements to be satisfied by DC/DC converters are high transformation ratios, high power density, high efficiency and low current and voltage ripple at the output. This paper presents the design and modeling of a buck-boost converter based on the concept of reduced redundant power processing with high voltage transformation ratio. The operating states of the converter during the switching process of the switches are presented. The average and linear models are presented, as well as some comparative aspects with some topologies reported in the literature.