Luiz Eduardo Sales e Silva

O sensoriamento remoto aplicado ao monitoramento de incêndios na Amazônia Brasileira tem passado por uma transição paradigmática com a introdução de métodos de inteligência artificial. Este trabalho apresenta uma revisão sistemática da literatura focada no desempenho de algoritmos de aprendizado de máquina (machine learning) e aprendizagem profunda (deep learning) para a modelagem de sistemas de predição de incêndio em florestas, com atenção particular à Floresta Amazônica. Foram analisadas arquiteturas de aprendizado por conjunto (Random Forest e XGBoost), redes neurais recorrentes (LSTM) e convolucionais (CNN), entre outras possibilidades, correlacionando-as com variáveis geoespaciais de clima, topografia, uso do solo, entre outras. Os resultados demonstram que algoritmos como o XGBoost superam métodos tradicionais em termos de erro residual em bases de dados de alta dimensionalidade. Identificou-se que a integração de redes LSTM permite a modelagem de dependências temporais críticas para a predição de risco sazonal, enquanto técnicas de análise espaço-temporal revelam a persistência de clusters de fogo em eixos antrópicos. A revisão destaca que a eficiência dos sistemas de monitoramento atuais reside na escolha adequada das variáveis a serem utilizadas no sistema, e da capacidade dessas tecnologias de processar grandes volumes de dados com alta acurácia, reduzindo falsos positivos e permitindo o mapeamento preventivo de áreas críticas no bioma amazônico.

The retrofitting of legacy systems enables upgrades that extend the lifespan of outdated equipment, improve efficiency, and reduce environmental impacts. This manuscript builds on existing approaches to retrofitting legacy systems using Industry 4.0 technologies. Therefore, it explores how the proposed modernization envisions the transition from Industry 4.0 to Industry 5.0, which emphasizes human-centric approaches, sustainability, and resilience. Tools such as RAMI 4.0 (a reference architecture model for Industry 4.0), Lean Six Sigma (a methodology for process improvement), and Big Data analytics are highlighted throughout the text as essential for optimizing processes and ensuring alignment with global challenges, including resource efficiency and environmental sustainability. This work addresses both conceptual and technical aspects of system modernization. It provides a comprehensive framework for retrofitting systems and integrating advanced technologies such as digital twins. These efforts ensure that industries are prepared for the evolving demands of Industry 4.0 and beyond.

Electric vehicles (EVs) are transforming the transportation sector, driven by the rapid expansion of charging infrastructure, including fast-charging stations (FCSs), significantly reducing charging time compared to standard charging stations. Despite the advantages of faster charging, the substantial power demand of EVs poses significant technical challenges for distribution networks. In particular, the existing literature has a research gap regarding how FCSs may impact or interact with step-voltage regulators’ (SVRs) tap operations. In this study, we characterize and evaluate the effects of fast recharging at varying penetration levels (PLs) on SVRs’ tap operations using probabilistic simulations and sensitivity analysis. To address these challenges, we propose a local and innovative application of the Volt/Var control on EV fast charging. The proposed application aims to inject reactive power into the network, depending on the FCS’s nominal active power, when the bus voltage connected to the FCS exceeds a minimum value. Our research on an actual feeder in northern Brazil reveals that reducing the active power supplied to the vehicle or oversizing the charging station power converters is unnecessary. Furthermore, our strategy reduces the probability of undervoltage violations and minimizes SVR tap changes, mitigating EVs’ impact on voltage quality.

In this paper, the practical application of automatic control techniques is performed to enhance the performance of the isochronous speed regulator of hydraulic turbines, thus making the electrical power system more efficient. To achieve this, the dynamic model of the 2kVA small-scale power system is employed for controller tuning. This system encompasses a synchronous generator and a coupled DC motor, with the hydraulic turbine dynamics incorporated into a microcontroller that replicates the behavior of larger power plants. For this purpose, a fractional-order PID controller (FOPID) is developed, given that this controller offers two additional degrees of freedom for design purposes. Therefore, in the design of the FOPID controller, a methodology based on analytical values of gain and phase margin is used. Additionally, a classical PID controller is tuned using the pole placement method to match the performance of both controllers. The aim of the study is to assess the effectiveness of the proposed methodology through comprehensive computational simulations and practical experimental tests, considering variations in the reference value and load, and qualitative analysis through temporal and quantitative analysis, using performance indices. The results demonstrate that the proposed methodology outperforms other approaches and confirms its effectiveness and flexibility. Thus, this article significantly contributes to the field of power generation system control, highlighting the importance of experimental validation to ensure performance in real-world scenarios.

This paper presents a comparative analysis between grid-side solutions and distributed generation (DG) side solutions for overvoltage mitigation in low-voltage distribution networks with penetration of photovoltaic (PV) generation. The analysis is based on calculating the percentage of consumer units with voltage transgression over a simulation day for each increment in the degree of DG penetration and each mitigation technique studied. Grid-side solutions consist of network reinforcement and permanently changing the tap of MV/LV transformers, as well as both techniques together. The solutions on the DG side correspond to the voltwatt control and volt-var control functions in PV inverters and battery energy storage systems (BESS). The results presents that for the simulated case, with the exception of the permanent tap switch, all solutions promote a decrease in the percentage of overvoltage caused in the consumer units, with the volt-watt control and the BESS, solutions by DG side, increasing the hosting capacity of the network by up to 400% for heavy load demand.

This paper presents a novel direct form to design a digital robust control using RST structure (i.e., name given because of the R, S and T polynomials computed) based on convex optimization such as Chebyshev sphere; this approach was applied to a DC-DC Buck converter. This methodology takes into account parametric uncertainties and a Chebyshev sphere constraint in order to ensure robust performance and stability of the system in the discrete domain. For this purpose, a mathematical model for the DC-DC Buck converter is presented when considering uncertainties in electrical variables, such as load resistance, inductance, capacitance, and source voltage variation, also to obtain the discrete model of the system by using the bilinear transformation. The proposed methodology is compared with two other approaches designed in a discrete domain: the classical pole placement and the robust methodology based on the Kharitonov theorem. Wide-ranging experiments are performed in order to evaluate the behavior of the control methodologies when the system is subject to parametric variations of the load resistance and voltage setpoint variation. The results show that the proposed methodology outperforms the other approaches in 90% of the tests and ensures robust stability and robust performance when the system is subjected to a parametric uncertainties family.

This work presents a novel framework for detection and isolation of single-phase low-impedance faults in electric power systems. The proposed methodology employs spectral analysis based on Fourier transform for fault features extraction. The extracted features are used as input of artificial neural networks that aims to identify the occurrence (detection) and the location (isolation) of each fault. The proposed framework is validated through experiments that are performed using the IEEE 14 Bus benchmark. The experiments shows that the proposed methodology is efficient in detecting at least 80% of simulated faults, presenting good accuracy and generalization ability. The experiments also allow to relate the efficiency of the proposed methodology to the fault duration and RNA parameters (number of neurons).