ABSTRACT. The ability of machine learning techniques to leverage data and process rich perceptual inputs (e.g., vision) makes them highly appealing for use in autonomous systems. However, the inclusion of data and machine learning in the control loop poses an important challenge: how can we guarantee the safety of such systems? To address these safety challenges, we present a controller synthesis technique based on the computation of reachable sets, using optimal control and game theory. We present new methods that leverage advances in physics-informed neural networks to compute reachable sets and safety controllers efficiently. These techniques are highly scalable to high-dimensional systems, enabling the learning of safe controllers for a wide array of autonomous systems. Furthermore, these methods allow us to quickly update safety assurances online, as new environment information is obtained during deployment. In the second part of the talk, we will present a toolbox of methods that use data-driven reachable sets to stress-test learning and vision-based controllers. By identifying safety-critical failures, these tools guide performance improvement while maintaining safety. Together these advances establish a continual safety assurance framework for learning-enabled autonomous systems, where safety considerations are integrated across various stages of the learning process, from initial design to deployment and ongoing system enhancement. Throughout the talk, we will illustrate these methods on various safety-critical autonomous systems, including autonomous aircrafts, autonomous driving, and quadcopters.

ABSTRACT. Agriculture is one of the industries with the lowest percentage of automation in its processes, partly due to the difficulties associated with establishing communications and control in isolated locations without access to the electrical grid or the internet. For this reason, this article presents the design, implementation, and validation of a communication and automatic control network for Internet of Things (IoT) devices, applicable to remote or difficult-to-access scenarios. The main communication protocols to be used in each section of this network are presented, efficiently combining LoRaWAN, MQTT, and HTTP. The physical devices required for its implementation are also presented. The minimum applications required to control and communicate with the system are shown, and functional examples of these are provided. Finally, a practical implementation of this communication system is presented, demonstrating its effectiveness through various tests that prove the feasibility of this remote control system.

ABSTRACT. This article presents two mathematical optimization models to solve problems related to designing and planning 5G wireless communications networks using millimeter wave frequency spectrum. Firstly, a fractional binary programming model is proposed to obtain an optimal performance efficiency ratio to maximize users over the whole latency. This allows the planning of 5G networks operating in the millimeter wave (mmWave) frequency spectrum. Starting from a set of restrictions in the first model, we obtain linear and quadratic models. In addition, the binary variables for a 5G network are defined considering a random distribution of the base stations in an area of 1 km2. Together with the above, we establish a constraint of the model that guarantees that at least each user is covered by one active base station at a predefined distance. The second proposed model is a quadratic model that allows maximizing coverage towards users, allowing an efficiency that can be established in a 5G network. This is achieved by excluding the linear restrictions applied to the first linear model. Our models assume the existence of line of sight (LOS) for links between users and base stations. Finally, we consider instances composed of 30 to 50 base stations, and 500 to 3000 users, using distances of 200 and 250ms. Our numerical results show that the proposed models can optimally solve all the instances in a short CPU time.

ABSTRACT. In recent years, electrical machines have been garnering increased attention due to their numerous advantages, leading to the evolution of various high-performance control techniques. Such control techniques require more complex processors to be properly executed. This article comprehensively reviews embedded systems technology applied to electrical machines. An extensive literature review has been conducted in different academic databases and suppliers web sites. This work contributes to the state of the art in embedded systems for power electronic converters and control systems for electrical drives.

ABSTRACT. Over 97.5 percent of Earth’s water is saline, with only 0.8 percent accessible as freshwater through aquifers, rivers, and lakes. Rapid population growth has increased the demand for clean water, resulting in higher pollution levels from industrial waste and agricultural runoff. This pollution leads to eutrophication, degrading aquatic ecosystems. Therefore, it is essential to monitor water quality to mitigate such effects. We have developed a low-cost and low-power IoT system for realtime turbidity monitoring. This system uses LoRa technology to transmit data collected by a high-capacity microcontroller to a database for graphical visualization. We tested the prototype for low energy consumption, extended communication range, and quality of turbidity measurement. The results demonstrate the system’s effectiveness in continuous, real-time monitoring. This system is ideal for widespread deployment across lakes for early warning detection and assessing water quality issues.

ABSTRACT. Automated systems play a key role in the development of early warning mechanisms, with the objective of preserving lives and securing regions susceptible to volcanic activity. The aim of this article is to develop intelligent algorithms based on Machine Learning for the multiclass classification of micro-earthquakes originated at Llaima volcano, including tectonic earthquakes, long-period events, tremors, and volcano-tectonic earthquakes. Our method encompasses preprocessing, processing, feature extraction, feature selection, and classification stages. During the classification, we employ machine learning algorithms, specifically Decision Trees (DT), k-Nearest Neighbors (k-NN), and Support Vector Machine (SVM). The evaluation of our system performance, is assessed through the Balanced Error Rate on test data, yields significant results: 0.12 for DT, 0.10 for k-NN, and 0.08 for SVM. SVM algorithm presents remarkably results when applied our methodology to the feature selected matrix, which considers 29 key features, this achievement results in accuracy approaching 96% and specificity of 98%.

ABSTRACT. In recent years, the application of the Internet of Things (IoT) in agriculture has gained traction, including wearable devices in precision livestock farming (PLF). This paper explores the use of reinforcement learning algorithms for energy management in wearable IoT devices with energy harvesting applied in PLF scenarios. These devices monitor animal health, welfare, and location. Ensuring energy efficiency is critical due to the autonomous operation expected from these devices, which are often powered by limited-capacity batteries, and sometimes integrate energy harvesting systems. This study implements Proximal Policy Optimization (PPO), Soft Actor-Critic (SAC), and Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithms to optimize energy consumption, creating an energy manager that adapts the duty cycle, frequency of data transmissions, and frequency of GPS executions based on weather conditions, orientation, and battery charge. Experiments demonstrate that reinforcement learning-based energy managers can effectively adapt consumption to different conditions. Using 10 000 episodes for training, the TD3 algorithm achieved the best overall performance. However, the PPO algorithm performed better during the autumn and winter seasons. This study contributes to future real-world implementation of reinforcement learning techniques for energy management in IoT devices for PLF.

ABSTRACT. Currently, telecommunications companies seek to optimize wireless communication networks. This work proposes a practical method for classifying the coverage of the Long Term Evolution (LTE) network based on measurements of the Reference Signal Received Power (RSRP) parameter. Initially, the RSRP is captured by an industrial router, and a Distributed Federated Learning (FL) approach based on The Federated Averaging (FedAvg) algorithm technique is applied. Three classes of coverage are established: “poor”, “good”, and “regular”, represented by classes 0, 1 and 2, respectively. Analyzing the influence of the RSRP probability distribution on the global model's accuracy, it is shown that a uniform distribution allows for achieving higher accuracy in fewer update rounds and shorter convergence times. This is in contrast to a non-IID distribution, where each random variable does not have the same probability distribution and is not all mutually independent. The results obtained confirm the potential of federated learning for coverage classification in LTE networks, preserving data privacy by not sharing it directly between devices.

ABSTRACT. Differential flatness theory has proven useful in providing efficient solutions to advanced control and state estimation problems. However, its application to differentially flat systems whose flat output vector is not directly measurable may face practical limitations. Recently, the concept of flat inputs has emerged as a promising approach to overcome these constraints. Nevertheless, to the best of our knowledge, existing studies with flat inputs have not considered the presence of unknown external disturbances in controller design. To address this gap, this paper proposes a derivative-free Kalman filter-based disturbance observer using flat inputs for differentially flat nonlinear systems with stable internal dynamics whose flat output vector is not a measurable variable. The effectiveness of the proposed scheme is numerically verified for the well-known quadruple-tank process.

ABSTRACT. Piezoelectric actuators exhibit hysteresis between the driving voltage and the displacement, impacting the positioning accuracy. This nonlinear behavior poses challenges for system identification and control. In this paper, we implement and test several filtering techniques to estimate the system states using a set of available noise-corrupted measurements. We present the estimation of the system states via numerical simulations and Monte Carlo analysis. These insights contribute to identifying the most effective state estimator for the nonlinear system model.

ABSTRACT. In this work we consider the preliminary analysis of cyber attacks, in the form of Denial-of-Service (DoS) attack, on a power constrained Networked Control System (NCS). For insight in the proposed analysis we limit the study to first order plant models, but this is without loss of generality in the context of a NCS, and more so, it is in line with a Networked Process Control System (NPCS) which call to focus only on the dominant dynamics of the process to be controlled. The DoS attack is modelled as a Bernoully process that can take place either at the sender or the receiver side of the communication channel. For the communication channel model we consider here the standard additive white noise (AWN) channel. For the proposed setup, we then analyze the stability of the resulting feedback loop, characterized in terms of the interference resulting from the DoS attack.

ABSTRACT. The main obstacle to implementing classic control projects over long distances is the high cost of installing and maintaining the infrastructure necessary to guarantee reliable communication, for instance, through wired networks. Wireless Sensor Networks (WSN) came as an alternative to the need for cabling. However, given the loss of information inherent to this type of transmission, to guarantee the stability of the closed-loop system, it is necessary to include several nodes (close to each other) throughout the communication network. In this context, this paper describes the possibility of performing remote control via WSN, aiming to improve the trade-off: good performance of the control system and reducing network expenses by allowing a decrease in the number of nodes. The proposed strategy consists of replacing the use of a classical control technique (linear quadratic regulator -- LQR) that requires full-reliable communication with a Markov model-based control technique, which provides theoretical guarantees of closed-loop system stability even when there is the possibility of packet loss (as is usual in WSN context). To illustrate the efficiency of the proposed approach, simulations of the dynamic behavior of a plant controlled through a wireless communication network are carried out using real statistical data obtained via experimental tests in a workbench emulation of a WSN with two nodes.

ABSTRACT. This paper addresses the stabilization problem of an inverted pendulum system. A Takagi-Sugeno fuzzy model by incorporating an intermediate state in terms of the angular position is developed. In order to stabilize the nonlinear inverted pendulum system, a Takagi-Sugeno fuzzy state feedback control is developed. The proposed control strategy shares the same membership functions of the proposed fuzzy model. The stabilization condition is formulated as a linear matrix inequality by utilizing a Lyapunov function. Finally, the benefits of our proposal are illustrated via numerical simulations.

ABSTRACT. Traditionally, the real-time implementation of Model Predictive Control (MPC) has been limited by processing and storage requirements. Recently, the idea of using Artificial Neural Networks (ANN) to approximate MPC control laws, including implementations on Field Programmable Gate Array (FPGA), has been explored. This work presents a complete design flow from software controller to hardware implementation, utilizing Keras and QKeras for ANN design and quantization and HLS4ML with the AMD-Xilinx Design Suite for FPGA implementation. The evaluation and analysis conducted provides insights into the trade-offs involved in the proposed workflow. Experimental results are validated on a PYNQ-Z1 board, achieving latencies of less than one microsecond in the case study, demonstrating a hardware precision comparable to traditional MPC methods.

ABSTRACT. This paper addresses the control problem of Aeropendulum systems. The Aeropendulum, known for its nonlinear dynamics, will be approximated as a Takagi-Sugeno (T-S) fuzzy system by introducing a new state variable. A parallel distributed compensation (PDC) controller will be employed to control the derived T-S fuzzy model, which will then be applied to the nonlinear system. Using the Lyapunov function and incorporating slack matrices, a stabilization condition will be established in a set of Linear Matrix Inequalities (LMIs). Finally, simulation results, utilizing Aeropendulum parameters, will demonstrate the effectiveness and validity of the proposed technique.

ABSTRACT. This paper presents a new predictive speed control algorithm for the permanent magnet synchronous motor (PMSM) drive using the model-free control theory. In the proposed methodology, the motor speed is directly controlled by the q-axis voltage reference signal instead of the conventional cascaded control of speed and q-axis current. On this basis, a second-order ultralocal model is defined, which relates the second derivative of the motor speed to the q-axis voltage. The unknown function of the ultralocal model is estimated using a second-order extended state observer (ESO). A comparative analysis is conducted through simulations among four different speed control methods, including the proposed method and the first-order model-free speed control for the non-cascaded control scheme, as well as the proportional-integral (PI) controller and model-free controller for the cascaded speed control scheme. According to the results, the proposed method achieves a faster dynamic response compared to the rest of the control approaches.

ABSTRACT. An efficient and robust technique is presented in this study to control the voltage and current of the inverters in Master-Slave Organized AC Smart Island. In the smart island, parametric uncertainty and external disturbances challenge voltage/frequency and current control of inverters-based renewable energy resources. It is consequently essential to adopt a stable and robust control approach. This study proposes an adaptive hybrid robust controller using a sliding mode controller (SMC) and backstepping controller. SMC can also be robust to many external disturbances and uncertainties. The backstepping sliding mode control (BSMC) structure includes an adaptive mechanism. The switching gain is adaptively updated in real time for the adaptive BSMC. The Lyapunov theory has been used to prove the closed-loop system's stability. Matlab/Simulink is used to implement the suggested control technique, and several case studies are analyzed to evaluate its effectiveness. According to simulations, the suggested controller maintains islanded microgrid stability by controlling voltage and current precisely regardless of external disturbances and uncertainty.

ABSTRACT. The increasing demand for Electric Vehicles (EVs) has underscored the need to develop charging infrastructure. Traditional fast charger topologies, such as the Dual Active Bridge, only support charging one vehicle at a time and lack flexibility in terms of modularity and scalability in voltage and power. Recently, Modular Multilevel Converters have been proposed as a potential solution to address these limitations in high-power EV charging applications. This paper presents a control strategy for a Series-Parallel Modular Multilevel Converter based EV charger. This strategy considers a priority-based charging strategy to redistribute power according to each EV priority and maximum power capability. Furthermore, it allows for bidirectional power flow, providing Vehicle-to-Grid capability. Simulation results are presented using a simplified series-parallel modulular multilevel converter model developed in PLECS to validate the performance of the proposed control strategy.

ABSTRACT. This study presents an analysis and controller design for dual-output single inductor DC-DC buck converter. We derive a linear model of the system and subsequently analyzing its properties using multivariable control techniques such as Relative Gain Array, Participation Matrix, and Singular Value Decomposition. Interactions among different system components are identified, and the most influential inputs are highlighted. Based on these findings, a Linear Quadratic Integral controller is proposed and compared with a decentralized Proportional-Integral controller present in the literature. Simulations results demonstrate a clear improvement in the performance using the proposed controller, indicating its effectiveness in addressing the cross-regulation problem in the converter. This study provides a significant contribution to the design and control of dual-output DC-DC converters with a single inductor.

ABSTRACT. The Modular Multilevel Converter (MMC) is an Alternating Current (AC) to Direct Current (DC) converter that has gained widespread use in High Voltage Direct Current transmission due to its modularity, efficiency, and controllability. Recently, MMCs in back-to-back configurations have been proposed for additional applications such as wind energy conversion systems and low-frequency AC transmission systems where direct AC-AC is required. In such applications, controlling MMC presents significant challenges because AC-AC operation can lead to increased voltage fluctuations in the converter. This paper describes the use of a decoupled control strategy for a Back-to-MMC operating in LFAC. The control strategy uses the circulating currents in each MMC to maintain acceptable ripple levels in the capacitor voltage of each cell. The effectiveness of the decoupled control strategy is validated using real-time simulation models implemented with the OPAL-RT real-time simulator.

ABSTRACT. Multiport converters (MPC) have gained considerable attention due to their applications in renewable energy systems. An example of MPC is the single inductor multi-input multi-output (SI-MIMO) converter, which is capable of integrating different power sources to feed multiple output ports with a simple topology. This article proposes to evaluate the performance of the SI-MIMO converter in a 30 kW system fed with two renewable energy sources, consisting of a wind turbine and a photovoltaic (PV) array, to feed three sets of two 5 kW parallel-connected electrolyzers. A control strategy with maximum power point tracking (MPPT) for the PV generator and power control for the wind turbine is proposed. The simulation results in PLECS software validate the functionality of the SI-MIMO converter and the proposed control strategy, achieving an efficiency of 98.06%.

ABSTRACT. Efficient energy distribution and precise voltage regulation are essential for green hydrogen production using renewable sources; this research presents a hybrid DC microgrid model with a grid-forming control strategy tailored for green hydrogen production. The model integrates power electronics and renewable energy sources, with the Buck-Boost versatile converter (BBVC) playing a vital role in dynamic and static energy management, enabling scalable and efficient power regulation. The study also explores the integration of a photovoltaic (PV) system, a battery energy storage system (B), and a supercapacitor energy storage system (SC), along with their respective modeling methods. Additionally, the model incorporates proton exchange membrane (PEM) electrolyzers for green hydrogen production, demonstrating its capability to effectively integrate and manage diverse energy sources. This model yields valuable insights into the interactions between different components and control strategies within a DC microgrid, ensuring optimal performance and reliability.

ABSTRACT. ABSTRACT. Power systems are becoming complex dynamic systems given the current and future penetration of sustainable energy. In this sense, there are important challenges in the actual implementation of control and protection schemes to maintain security of suppy. This pannel will show actual industrial experiences from research labs and companies to project on the complex technical future of carbon-neutral power systems.

ABSTRACT. This paper addresses the optimal planning of the electrical grid in the Magallanes Region, focusing on integrating wind farms and distributing energy to industrial facilities for green hydrogen production. A mathematical optimization model based on linear power flow and genetic algorithms was developed to establish the transmission network's topology. Considering geographical, environmental, and power data, the study offers a robust methodology for sustainable planning of electrical systems in remote areas with renewable resources.

ABSTRACT. Isolated microgrids are essential stand-alone systems for addressing energy needs in remote areas. This research work applies a methodology for analyzing generation alternatives for a particular isolated area, which seeks to improve the current conventional generation alternative based on fossil fuels. Los Tres Gigantes Biological Station, located in the city of Bahia Negra in the Department of Alto Paraguay, in the Paraguayan Chaco, is presented as a case study. The methodology covers a stage of data collection, study of the Station's load curve, analysis of energy resources available in the area and their feasibility of use, sizing of electricity generation alternatives considering technical, financial and environmental criteria. The objective of this work is to propose current alternatives for electricity generation for Los Tres Gigantes Station, evaluating various renewable generation options, such as photovoltaic solar energy with different variants of energy storage, wind generation and biogas generation. For the study of biogas production, the Technology Readiness Levels (TRL) model has been used, in order to evaluate its viability, in which the aquatic species Eichhornia Crassipes (biomass) has been considered as the primary source, which exists in abundance in the zone. On the other hand, the HOMER Pro and PVSyst software have been used for simulations of solar, biogas and fossil fuel generation alternatives (this alternative for the comparison of energy and economic savings), according to the aforementioned criteria. The results obtained are positive and demonstrate the advantages and disadvantages of each of the generation alternatives proposed as a solution for the isolated area, considering the energy resources available in this area of Paraguay, the Chaco.

ABSTRACT. This article describes the development of a program in Matlab to automate the calculations of operating costs of fleets of electric buses as a study prior to the replacement of a fleet of diesel buses. The program estimates the electrical consumption of buses on specific itineraries and optimizes load management, allowing compliance with the regular operating program with the lowest possible operating cost. This is achieved through the adaptation and processing of real data provided by a diesel bus company. A program was developed in Fico Xpress that uses ANDE's tariff schedule to reduce power and energy contracting costs. The itinerary measurements made with GPS, previous data processing and physical modeling of an electric bus in Simulink make it possible to predict the energy consumption of the bus and determine the operational feasibility in the study itinerary for a typical week of operation. Before converting a fleet from diesel buses to electric, it is crucial to know this information to predict whether the specific bus model meets the necessary technical requirements. Furthermore, since it is an automated program, it is enough to alter some input parameters to analyze the performance of various bus models or itineraries and determine the minimum operating cost for each configuration.

ABSTRACT. In recent years, there has been a significant increase in the production of green hydrogen using renewable energy sources, particularly photovoltaic energy. This is seen as an effective way to combat the impacts of climate change. An electrical system is needed to power the electrolyzer continuously; to achieve this, the electrolyzer is connected with power converters and energy storage systems such as batteries, supercapacitors, and hydrogen tanks. These systems allow the hydrogen produced to be converted into electrical energy through a fuel cell when needed, ensuring continuous operation even when the renewable energy source is unavailable. This work proposes an optimization model for an off-grid DC microgrid aimed at maximizing the utilization rate of the electrolyzer and minimizing hydrogen fuel cell consumption. The results demonstrate that a linear method can reduce hydrogen fuel consumption by approximately 40.73 % to meet electrical demand while increasing hydrogen production and optimizing energy generation and storage systems.

ABSTRACT. Energy management of buildings is crucial to cope with increasing energy demand and reduce environmental impact. This article presents a review of the implementation of ISO 50001 in buildings around the world, with a focus on energy efficiency and sustainability. Through a structured methodology that includes a literature review and data analysis, the study examines the current status of ISO 50001 adoption in buildings, including public and university facilities. Key research questions are addressed, such as global leaders in ISO 50001 adoption and the financial implications of implementing energy-saving measures. The results highlight successful case studies, which demonstrate significant energy and emissions reductions through the implementation of ISO 50001. In addition, the study emphasizes the need to increase ISO 50001 certifications in public and university buildings, especially in Latin America, to promote energy efficiency and reduce dependence on non-renewable energy sources. The findings provide valuable information for future research and advocate the widespread adoption of ISO 50001 to improve energy management practices in buildings worldwide.

ABSTRACT. In recent years, Chile has become a focus of study for green hydrogen projects due to its high potential for renewable energy and its commitment to decarbonizing the energy matrix through already established public policies. This document presents an optimization model for the transportation of green hydrogen through pipelines using the simulation of a hydrogen production system model in the Antofagasta region, specifically in the Atacama Desert. The results show that the best option for transporting hydrogen from the production plant to the export port is through pipelines, which should be sized to a diameter of 0.15 m.

ABSTRACT. This article presents a novel approach for modeling water distribution networks (WDNs) based on physical principles, graph theory, and the pseudo-transient continuation method. The main feature of this approach is that permits the representation of WDNs as fully interconnected networks of damped nonlinear oscillators, with each oscillator formulated as a Liénard system given in terms of the flow rates. This representation facilitates the design of control and observation algorithms, particularly in scenarios where the WDNs lack pressure sensors. To show the possible usages of this approach, a state observer for estimating the user's water demands is designed and numerically tested.

ABSTRACT. This work contains the data analysis of advanced process control systems implemented in the atmospheric distillation columns of Aconcagua refinery, Empresa Nacional del Petróleo. The main problem in production control of atmospheric distillation and in other types of processes is the difficulty to continuously measure all relevant parameters in real time, as for example the quality of the final product according to laboratory reports. Moreover, outliers that appear in operational databases affect the accurate interpretation of production results. It is important to address these problems, since the implementation of quality control methods and having reliable data allows reducing opportunity costs and downtimes by anticipating potential problems before they significantly affect production. By using the statistical software R, it was possible to filter outliers from operational databases, perform trend analysis, elaborate graphs to compare different control modes and estimate economic benefits associated with the use of advanced controllers of quality, performance and resources related to energy consumption. Therefore, it was possible to quantify the advantages from advanced control systems related to diesel and kerosene production, as well as the reduction of fuel gas consumption from preheating furnaces and the reduction of steam used in the operation of kerosene strippers and the bottom of the distillation columns. As future work, it is considered that the developed data filtration and analysis will not only benefit the monitoring of the atmospheric distillation process, yet it can also be extended to analyze other production processes.

ABSTRACT. This paper describes the implementation of an advanced artificial intelligence-based system for the detection and labeling of feed sacks, using the Jetson Nano development board and the YOLOv8 neural network model, this system employs computer vision techniques to accurately identify the sacks and their side labels. Ensuring the pre-presence of a label on each sack is crucial for product monitoring and quality control of the labeling process. The methodology involves a rigorous model training process, using data captured in real production conditions, and system configuration to optimize detection performance. The results obtained demonstrate the system's robustness and efficiency, with an accuracy rate of 98.84% in controlled conditions and 96.42% in a real production environment. This underscores the system's practicality and its potential to significantly enhance logistics and quality control processes in the shrimp feed industry. The technology's implementation can lead to improved product traceability and operational efficiency throughout the supply chain, potentially reducing costs, minimizing errors, and increasing end-customer satisfaction.

ABSTRACT. This work complements other previous studies and developments carried out by the authors on specific applications of UAVs at an industrial level for the operation of a fleet of drones within a production plant characterized by the customization of products in a flexible production system of large productive volumes. In particular, the problem of determining the drone fleet for this type of flexible and personalized production plant that allows demand to be met is addressed. A Case Study developed in a Plastic Manufacturing Company is presented, through a drone fleet management model based on energy efficiency, and its implementation according to annual, monthly, daily and shift planning. Different scenarios are considered in estimating the size of the drone fleet, including the variation in demand for final products, variability of critical process parameters, load coefficient for different demands and availability of the drone and the fleet.

ABSTRACT. As the global automotive industry transitions to sustainable transportation, electric vehicles are expected to be the replacement for vehicles with internal combustion engines. To support the massification of EVs, Electric Vehicle Supply Equipments (EVSEs) will be required. This paper presents the design and implementation of an EVSE designed for alternating current charging. The proposed EVSE is composed of a real-time control system and control systems to provide user-friendly charging, targeting an advanced technology research level. Experimental results obtained from the charging of a commercial electric vehicle are presented demonstrating the effectiveness of the proposed EVSE in real-world conditions.

ABSTRACT. This study examines the implementation of embedded systems and advanced power converters in the generation of green hydrogen through solar energy. Various topologies such as interleaved buck and multilevel converters are analysed, highlighting their role in optimizing the efficiency and stability of the system. The integration of Hardware-in-the-Lopp (HIL) technologies such as National Instruments and dSPACE allows advanced control systems to be validated and developed. This work highlights the potential of these advances to improve energy sustainability and promote cleaner, more efficient infrastructure.

ABSTRACT. This paper presents an improved sliding mode-based finite control set model predictive controller for grid-tied single-phase multilevel inverter systems with an LCL filter. The proposed controller employs the inherent robustness of sliding mode control (SMC) with the fast dynamic response of the FCS-MPC method for controlling the complex behavior of LCL filter-based grid-tied inverter systems. The LCL-based grid-tied multilevel inverters necessitate the control of inverter side and grid side currents in addition to the grid filter capacitor voltage and DC-link capacitors voltage control/balance. Incorporating FCS-MPC with SMC eliminates the chattering effects in the traditional SMC algorithms. Moreover, incorporating SMC with FCS-MPC enables more robust and stable control performance compared to the conventional FCS-MPC method. The proposed method eliminates the need for weighting factors for DC-link capacitor voltage balance by properly selecting redundant voltage vectors. Simulation results are provided to validate the proposed SMC-based FCS-MPC controller with different active and reactive power injections to the grid.

ABSTRACT. There is a growing trend towards the mass adoption of Electric Vehicles (EVs), driving the need for advanced EV charging technology. EV chargers cover various modes and power ratios with the primary goal of reducing charging times. Fast electric vehicle chargers typically use direct current and require galvanic isolation, requiring multiple conversion stages that increase cost and complexity. Recently, transformerless DC chargers have been proposed as a cost-effective alternative that simplifies EV chargers by reducing the number of conversion stages. However, there are few articles on this topic. This paper provides a comprehensive overview of electric vehicle chargers, with a particular focus on fast charging technology, standards, and regulations. Special attention is given to transformerless technology, describing topologies and key control strategies to highlight its potential advantages in EV charging applications.

ABSTRACT. The Modular Multilevel Matrix Converter (M3C) has been proposed for several high-power applications due to its numerous advantages in terms of power quality, efficiency, controllability, and fault tolerance. In particular, the M3C is a promising alternative for direct alternating current applications such as motor drives, wind power, and low-frequency AC transmission. Despite being indicated as a fault-tolerant converter, only some articles are available in the literature on this topic. Therefore, this paper proposes an M3C basic control system for fault tolerance during cell failures. The proposed control system is based on reordering the balancing control of a cluster when power cells experience faults, ensuring proper operation and keeping the power transference capability. Simulation results obtained using PLECS software are presented.

ABSTRACT. In this work, the implementation of a Model Predictive Control (MPC) with circular constraints applied to a Distributed Energy Resource (DER) connected to the grid via an inductive coupling is presented. The main goal is to observe how circular constraints perform in tracking currents in dq coordinates. For this purpose, Matlab Simulink will be used along with fmincon as the solver, which is well-known for its robustness in handling nonlinear constraints.

ABSTRACT. In this paper, a primary model-based predictive control (MPC) for Distributed Energy Resources (DERs) is studied. This primary control regulates voltage and frequency, and has the capacity to set a predefined operation point while simultaneously imposing security constraints on the inverter's output current. To achieve these objectives, the MPC takes into account the LCL filter dynamics to increase the prediction accuracy of the system. The MPC algorithm is formulated as a Quadratic Programming (QP) problem which is solved using a first-order optimization method. To validate this approach, simulation results using MATLAB/Simulink and Plecs are presented. The results confirm the efficacy of the proposed primary control method.

ABSTRACT. This work introduces a current control strategy for Voltage Source Inverters (VSI) using data-driven control systems, particularly employing a framework based on Deep Reinforcement Learning agents. Unlike the other techniques in the literature, we have avoided using a modulator by including a Deep Q-Network agent. In addition, an analysis of the impact of different Deep Neural Network (DNN) architectures on control system performance, specifically considering the number of layers and neurons, is presented. To this end, different DQN agents were designed, trained, and tested. Also, a two-level voltage source power inverter is simulated to validate the proposed data-driven control based on DQN agents. The performance of the control strategy is analyzed in terms of computational cost, root mean square error (RMSE), and total harmonic distortion (THD). Simulated results reveal that the proposed control strategy performs strongly in the current control, with a maximum RMSE of 0.83 A and a THD of 5.29% at a 10 kHz sampling frequency when a DNN with one layer and five neurons is used.