EMC - Trabalhos de Conclusão de Curso

URI Permanente para esta coleçãohttp://200.137.215.59//handle/ri/1309

Navegar

Agora exibindo 1 - 20 de 168

Filtragem de ruído em sinais de eletrocardiograma e áudio utilizando autoencoders e mecanismos de monitoramento
(Universidade Federal de Goiás, 2025-06-25) Moreira, Isaías Ramos; Cardoso, Álisson Assis; Cardoso, Álisson Assis; Lemos, Rofrigo Pinto; Corrêa, Henrique Pires
This study proposes an approach based on autoencoders with residual blocks and monitoring mechanisms for filtering electrocardiogram signals contaminated by noise, both in isolation and in combination, as well as spectrograms corrupted by common urban environmental noise generated from audio files. Two distinct architectures were evaluated: one using a Squeeze-and-Excitation block and another incorporating a monitoring mechanism called the Attention Block. The models were tested on electrocardiogram signals and spectrograms contaminated with noise to assess their performance in signal reconstruction and noise removal tasks. The Attention Block demonstrated superior performance in both tasks. This advantage is attributed to its ability to integrate channel-wise attention mechanisms and local noise detection, which proved relevant when multiple types of i
Adequação do método da soma das potências versão retangular considerando elementos shunts de barra e de ramos de redes de distribuição
(Universidade Federal de Goiás, 2025-06-25) Santos, Ronny Clézio Dias dos; Brigatto, Gelson Antônio Andrêa; Brigatto, Gelson Antônio Andrêa; Garces Negrete, Lina Paola; Kopcak, Igor
Electrical distribution networks are mostly composed of series elements, which justifies the use of classical power flow algorithms based on this topology, such as the Power Summation Method (PSM) in its polar and rectangular versions. However, the inclusion of shunt elements—such as voltage regulators, capacitor banks, reactors, and the capacitive effects of feeders—requires adaptations to traditional methods. While the formulation of the PSM in polar coordinates with shunt elements is already discussed in the literature, there is still a lack of equivalent development for the PSM in rectangular coordinates. This work aims to propose a formulation that incorporates these elements into the rectangular PSM. The necessary equations are developed, followed by computational tests to evaluate the accuracy and performance of the proposed method compared to existing approaches. The results indicate that the new formulation preserves the reliability of classical methods while extending their applicability to more complex network configurations.
DigSILENT PowerFactory software’s potential for analyzing electrical power systems
(Universidade Federal de Goiás, 2025-06-30) Amaral, Anna Beatriz Fernandes; Garces Negrete, Lina Paola; Garces Negrete, Lina Paola; Brigatto, Gelson Antônio Andrêa; Kopcak, Igor
Modern power systems require rigorous analysis to meet operational criteria of stability, efficiency, and compliance with operational standards. This research investigates steady- state load flow analysis using DigSILENT PowerFactory software, employing two case studies to simulate power flow through the Newton-Raphson method. The main results evaluated correspond to voltage magnitudes, phase angles, and active/reactive power distribution, identifying voltage deviations from the American National Standards Institute (ANSI C84.1- 2020) voltage standards. Tests performed on the IEEE 14-bus system and Peru’s National Interconnected Electric System (SEIN) yielded results demonstrating efficient convergence in few iterations, validating the algorithm’s robustness for both small and large-scale networks, while its computational accuracy was confirmed through comparative analysis with MATLAB simulations. The software's integrated visualization tools and automated reporting features highlight its usefulness for analyzing voltage profiles and power flows, while emphasizing the importance of voltage regulation strategies and reactive power management to fulfill operational constraints. Beyond the theoretical study, the research extended to simulating Peru’s National Interconnected Electric System (SEIN), demonstrating PowerFactory’s capability for large-scale simulations that, in this case, revealed critical challenges such as overvoltages at specific buses and transmission losses, reinforcing the need for optimizations in planning and operating complex power networks.
Estratégias para a mitigação de fluxo reverso em redes de distribuição com alta inserção de geração solar fotovoltaica
(Universidade Federal de Goiás, 2025-07-02) Pereira, Ivan Santos; Vergara, Gustavo da Costa; Garces Negrete, Lina Paola; Garces Negrete, Lina Paola; Vergara, Gustavo da Costa; Belchior, Fernando Nunes; Kopcak, Igor
The electric power sector has undergone significant changes in recent times due to the growing demand for a cleaner and more sustainable energy matrix. In line with this trend, the Brazilian government began encouraging energy generation by consumers themselves through the Legal Framework for Distributed Generation (Law No. 14,300/2022 from Brazil), which has made photovoltaic distributed generation a rising trend in Brazil. However, the expansion of distributed generation (DG) can lead to negative impacts on the grid, such as reverse power flow and overvoltage at system buses. This situation raises concerns for utility companies and motivates studies aimed at monitoring and mitigating these adverse effects. This study aims to propose strategies to mitigate reverse power flow in distribution networks with high photovoltaic generation penetration and to assess their impact on the electrical system. The research proposes three strategies for mitigating reverse flow: a battery energy storage system (BESS), an inverter-based Grid Zero control, and PV Curtailment, which consists of limiting the power injected by the modules with an inverter when an increase in bus voltage is detected. The proposed strategies were implemented in a distribution network in São Paulo using OpenDSS, an open-source software developed by Electric Power Research Institute (EPRI) that performs power flow analysis. The assessment of effects included the analysis of power flow through critical transformers, the substation, as well as the monitoring of bus voltages and network energy losses. The results demonstrate that the proposed strategies were effective in mitigating reverse power flow. Battery allocation yielded the best results, enabling the full use of energy generated by photovoltaic plants. The Grid Zero inverter curtailed part of the generation but was effective in mitigating reverse flow and presents a more economical alternative. PV Curtailment was the least effective in attenuating reverse flow, but managed to decrease the voltage rises in the nodes more efficiently than the other cases did.
Aplicação de otimização por enxame de partículas no projeto de linhas de transmissão aéreas de extra alta tensão
(Universidade Federal de Goiás, 2025-06-30) Barbalho, Aliucha Morais; Jesus, Ana Gabriela Machado de; Kopcak, Igor; Garces Negrete, Lina Paola; Brigatto, Gelson Antônio Andrêa; Kopcak, Igor
This paper presents the implementation of the Particle Swarm Optimization (PSO) algorithm in the design of transmission lines. The main objective is to highlight the benefits of using the algorithm to navigate a complex scenario in the evaluation of the expansion of extra-high voltage overhead lines, finding the best alternative for cost reduction, considering a set of possibilities for the number of conductors per bundle, variation in the cross-section of the conductors and spacing of the geometric arrangement of the bundle. The applicability of the proposed tool was motivated by two case studies for transmission line scenarios in countries with substantial territorial extension, whose results proved its viability as a support for the design of transmission lines. The application of the (PSO) proved to be effective in the optimization of design variables, such as gauge, number of subconductors and spacing, allowing the identification of configurations that balance costs and electrical performance. The developed tool allows the structured recording of input parameters, the documentation of the results obtained and the comparative analysis between alternatives. Thus, it contributes to the technical-economic planning of new lines, allowing for the reduction of losses, maximization of transmission capacity and greater efficiency in the use of resources. Thus, the use of meta-heuristic methods such as PSO represents a promising alternative to assist engineers and planners in decision-making in electric power transmission systems.
Adequação do método da soma das correntes considerando elementos shunts de barra e de ramos de redes de distribuição
(Universidade Federal de Goiás, 2025-06-25) Oliveira, Arthur Custódio de; Rocha Júnior, Enes Maximino da; Brigatto, Gelson Antônio Andrêa; Brigatto, Gelson Antônio Andrêa; Garces Negrete, Lina Paola; Kopcak, Igor
The modeling of electrical power distribution networks usually considers the network topology as consisting only of series branch elements. This characteristic underlies the classical algorithms for calculating the load flow of distribution networks, such as the Current Sum Method (CSM), so that the inclusion of shunts such as voltage regulators, capacitor banks or reactors and capacitive effects of feeders requires adaptations to the traditional methods. Although the formulation of the Power Sum Method considering shunts has already been addressed in the literature, the equivalent for CSM needs to be developed. This work aims to propose an adaptation of the Current Sum Method for load flow analysis in distribution networks, with the inclusion of reactive shunt elements of buses and branches of the network. The work contemplates the reformulation of the classic MSI with the incorporation of new terms to the equations, as well as the application of the computational implementation of the proposal in test networks for performance comparison and validation of the results with already consolidated solution methods. Case analysis demonstrates that the new approach allows representing the effects of reactive elements on voltage profiles and system losses equally accurately. Thus, this work contributes with a model applicable to the simulation requirements of actual distribution networks, with potential for adaptation to scenarios with distributed generation and control devices.
Otimização de pré-codificação conjunta com formação de clusters em um sistema cell-free auxiliado por RIS
(Universidade Federal de Goiás, 2025-06-23) Tealdi, Paulo; Coelho, André Almeida Souza; Lemos, Rodrigo Pinto; Lemos, Rodrigo Pinto; Cardoso, Alisson Assis; Castro, Marcelo Stehling de; Coelho, André Almeida Souza
This final course project aims to present the theoretical foundations related to Reconfigurable Intelligent Surfaces (RIS) and the cell-free architecture, including construction aspects, phase-shift constraints of the RIS, performance analysis, and comparisons. Based on this, a clustering algorithm for RIS-aided cell-free networks is proposed, in which, for each user equipment (UE), a set of access points (APs) responsible for transmission is dynamically selected, as well as a subset of RISs that will assist in communication. In this way, cluster formation is carried out in a UE- centric manner, aiming to optimize the WSR (weighted sum-rate) and reduce signaling overhead. The optimization of the WSR with respect to the active precoding vector (precoding at the APs) and the passive precoding (RIS phase-shift matrix) is performed based on existing alternating optimization algorithms in the literature. Therefore, with the subdivided problem, the lustering part is introduced and solved here using a greedy algorithm.
Implementação da ITIL e COBIT como estratégias para melhoria da gestão de serviços de TI
(Universidade Federal de Goiás, 2024-12-12) Ribeiro, Pedro Henrique Fernandes; Castro, Marcelo Stehling de; Castro, Marcelo Stehling de; Santana, Adriano César; Oliveira, Gustavo Dias de
This paper aims to implement ITIL and COBIT strategies to improve IT Service Management in a fictional company. The research includes a detailed analysis of the company’s IT infrastructure, identifying gaps in systems, processes, and security controls, as well as developing a plan for the implementation of ITIL and COBIT best practices. The goal is to align IT with the company’s strategic objectives, promoting increased security, operational efficiency, and IT governance. The methodology involves a literature review on ITIL and COBIT, analysis of the IT infrastructure, identification of improvement areas, and the creation of an implementation plan. IT governance is treated as a set of practices that ensures alignment of IT with organizational objectives, along with continuous performance monitoring. The application of ITIL and COBIT frameworks aims to optimize IT management by minimizing risks and enhancing organizational efficiency. The implementation was organized in phases such as analysis, planning, execution, testing, and continuous monitoring, ensuring compliance and effectiveness of the changes. Additionally, Google Colab and Python were used to create graphs and analyze data related to the implementation, utilizing the Matplotlib library to generate performance charts and success indicators. These graphs helped visualize the evolution of IT processes and identify areas requiring adjustments, facilitating strategic decision-making. The experience gained can serve as a model for other companies looking to improve their IT governance and operational efficiency.
Simulações computacionais de escoamentos sobre aerofólios e asas usando software livre
(Universidade Federal de Goiás, 2025-12-11) Bastos, Isadora Venancio; Nascimento, Andreia Aoyagui; Mariano, Felipe Pamplona; Mariano, Felipe Pamplona; Beghelli, Júlio Modesto; Nascimento, Andreia Aoyagui
Wings operate with maximum efficiency when they achieve the highest ratio between the lift coefficient and the drag coefficient. These coefficients are dimensionless numbers related to the interactions exerted by a fluid flow on solid objects and are essential for optimizing wing designs as well as for other aerodynamic applications, such as optimizing wind turbines, compressors and hydraulic pumps. The computational approach makes it possible to develop different studies for different aerodynamic profiles at little cost compared to experiments, so it is possible to analyze in detail how a wing behaves under different flow conditions. In order to investigate these coefficients, a geometric model and three-dimensional complex meshes were developed to study air flow over a wing formed by a NACA 4415 profile, with a 10º angle of attack, unit chord length and two different spam lengths. This analysis is conducted with a Reynolds number of 750,000. For this purpose, the OpenFOAM software was chosen, as it is a free, open-source software dedicated to solving Computational Fluid Dynamics (CFD) problems. In addition to OpenFOAM, CAD software was used to generate the geometric model, in this case the wing; the SnappyHexMesh mesh generator, supplied with OpenFOAM itself, and to choose, among those available, the most suitable computational algorithms to carry out the simulations. Specifically, SimpleFOAM was chosen, a classic algorithm that solves the Navier Stokes equations in steady state for incompressible and turbulent flow. The main result obtained is the development of a specific process for carrying out computer simulations of three-dimensional turbulent flows over complex (non-Cartesian) geometries.
Comparação de arquiteturas de redes neurais convolucionais para a detecção de doenças foliares do tomateiro
(Universidade Federal de Goiás, 2024-12-20) Lindolfo, Glauber Borges; Chaves, Ian Marcos da Cruz; Vinhal, Cassio Dener Noronha; Vinhal, Cassio Dener Noronha; Cruz Junior, Gelson da; Oliveira, Marco Antonio Assfalk de; Rocha, Flávio Geraldo Coelho
The automated recognition of leaf diseases is one of the main challenges of Agriculture 4.0, requiring methodologies that integrate agronomic knowledge, image collection for monitoring, and advanced machine learning techniques. This study aims to perform a comparative analysis of different convolutional neural network (CNN) architectures applied to the detection of leaf diseases in tomato plants, including target spot. Three widely recognized architectures were used: ResNet-50, Inceptionv3, and VGG-16, exploring combinations of hyperparameters such as learning rate, optimizers, and the use of weight decay. Additionally, activation maps were employed to identify relevant visual patterns that influence the models’ decisions. The results show that ResNet-50 achieved the best overall performance and stability, followed by Inception-v3, while VGG-16 exhibited greater sensitivity to training configurations. Through this analysis, we aim to understand the impact of these variations on network performance, providing valuable insights for improving models applied to crop protection management and precision agriculture.
Análise de custos de manutenção preventiva/preditiva de reparos dos cilindros de empilhadeiras comparado com parada não programada: estudo de caso
(Universidade Federal de Goiás, 2024-12-16) Souza, Pedro Victor Oliveira; Figueiredo, Kléber Mendes de; Figueiredo, Kléber Mendes de; Fonsceca, João Paulo da Silva; Oliveira, Ademyr Goncalves de
This work aims to conduct a cost analysis of replacing the set of seals for the forklift cylinders of a forklift rental company. The compared values refer to the current corrective maintenance model and the proposed preventive/predictive maintenance model, analyzing the costs of labor hours, transportation, and materials for maintenance. The results showed a 30% cost reduction when using the preventive maintenance strategy instead of corrective maintenance, in addition to determining the practicality of implementing preventive maintenance over predictive maintenance based on the business nature of the company under study.
Implementação e integração de uma arquitetura de software para um crawler de postagens em plataformas digitais
(Universidade Federal de Goiás, 2024-12-19) Cesar, Lucas Rezende Soares; Brenner, Thor Franco; Barbosa, Jacson Rodrigues; Barbosa, Jacson Rodrigues; Oliveira, Marco Antonio Assfalk de; Graciano Neto, Valdemar Vicente
The study addresses the issue of misinformation on social networks, a phenomenon exacerbated by artificial intelligence technologies and mass dissemination strategies, which undermine the credibility of democratic processes and the formation of public opinion. The research proposes a technical solution to automate the collection and analysis of data from fact-checking agencies and digital platforms, employing crawlers and APIs, with a focus on integration with the Web 3.0 Project. The results demonstrated the effectiveness of the developed architecture in consolidating structured data from reliable sources and social networks, enabling analyses such as sentiment analysis to identify polarizations and social trends. Despite limitations imposed by social network APIs, the system proved scalable and functional, contributing to a more agile and accessible fight against misinformation. Thus, the study highlights the importance of integrating technology and human expertise to tackle complex informational challenges.
Análise energética em uma instalação elétrica industrial
(Universidade Federal de Goiás, 2024-12-05) Pacheco Junior, Paulo de Souza; Oliveira, Antônio Melo de; Belchior, Fernando Nunes; Belchior, Fernando Nunes; Mariano, Felipe Pamplona; Moreira, Leonardo de Queiroz
This study investigates energy efficiency in an industrial electrical installation, focusing on power quality and optimizing the compressed air system, one of the largest energy consumers in industrial environments. The research explores practical and theoretically grounded measures to reduce energy consumption, enhance operational efficiency, and ensure the plant's sustainable growth—factors aligned with the United Nations' Sustainable Development Goals. Key approaches include reducing operating and cut-off pressure, recovering heat from compression, and controlling leaks. Efficiency analysis is conducted through methods such as plant consumption profiling, energy demand diagnostics, and calculating specific indicators like energy consumption per unit and system performance. Economic indicators, including Net Present Value (NPV), Internal Rate of Return (IRR), and simple payback period, were used to compassess the financial feasibility and savings potential of the proposed improvements. Additionally, the study highlights the energy-saving potential of recovering compression heat for preheating water used in other industrial processes, demonstrating reductions in both electricity consumption and boiler fuel costs. The results emphasize the strategic importance of energy efficiency for energy-intensive industries, showing that the applied measures contribute to more sustainable management and significantly reduce operational costs.
Análise do consumo de energia elétrica em grandes consumidores
(Universidade Federal de Goiás, 2024-12-09) Reis, Amanda de Sousa Batista; Cruz, Daniel do Prado Mendes; Belchior, Fernando Nunes; Castro, Marcelo Stehling de; Castro, Marcelo Stehling de; Valle, Ana Cláudia Marques do; Belchior, Fernando Nunes
This study analyzes the consumption profile of a Consumer Unit (CU) classified under Group A, supplied by Equatorial Goiás. The CU, part of the hotel industry, is characterized by high energy consumption, operates under the “Azul” tariff modality, and benefits from the Electric Power Compensation System (SCEE), which fully offsets its energy consumption. The analysis was conducted using Business Intelligence tools and methodologies for data processing, modeling, and visualization, supporting result interpretation and informed decision-making. The study examines current resolutions and tariff structures, as well as succinctly discusses the potential for leveraging the available data and information.
Diagnóstico de qualidade de energia elétrica em empresa com usina solar fotovoltaica
(Universidade Federal de Goiás, 2024-12-05) Pimenta, Jader Fillipe Cabral; Belchior, Fernando Nunes; Belchior, Fernando Nunes; Santos, Josephy Dias; Oliveira, Antonio Melo de
Electricity is an indispensable resource in modern society, essential for economic, social, and technological development. Its importance lies in sustaining the operation of basic infrastructure, such as hospitals, transportation, telecommunications, and supply systems, permeating all areas of daily life, from household activities to industrial processes. However, mere access to energy is not sufficient; the quality of supply is equally crucial. Parameters of Power Quality (PQ), such as voltage and frequency stability, harmonic control, and power factor, directly influence the reliability and efficiency of the supply. Non-compliance with these standards can result in power outages, equipment overloads, and damage, leading to economic losses and safety risks. Thus, maintaining high PQ is essential to prevent interruptions and ensure user well-being. In the industrial sector, the importance of PQ is even greater due to the complexity of production processes. High-quality electricity is critical for the efficient operation of precision equipment, automated systems, and production lines, as variations such as overvoltage’s, harmonic distortions, and frequency fluctuations can cause unexpected stoppages, shorten equipment lifespan, and increase maintenance costs. The electricity photovoltaic generation has been widely adopted in industrial facilities as a cost-reduction strategy and for being a more sustainable source in terms of environmental preservation. This practice aligns with the Sustainable Development Goals (SDGs) established by the United Nations (UN), particularly SDG 7, which aims to ensure universal access to affordable, reliable, sustainable, and modern energy. This study analyzed the relationship between this type of generation and the compliance with PQ parameters in an industrial facility. Concepts were addressed, and aspects such as steady-state voltage, power factor, harmonic distortions, frequency, voltage imbalance, and reverse power flow were evaluated. Real data collected over a week were graphically processed to illustrate the impacts of power injection from the plant on quality indicators. Compliance was assessed based on PRODIST standards.
Proposta de otimização do plano de manutenção de uma autoclave utilizada em ambientes hospitalares
(Universidade Federal de Goiás, 2024-11-11) Pinto, Carlos Daniel Silva; Figueiredo, Kléber Mendes de; Figueiredo, Kléber Mendes de; Vilela, Carlos Alberto de Almeida; Colvero, Diogo Appel
The study was motivated by the critical importance of sterilization performed by autoclaves for the proper and safe treatment of patients in hospital environments. It aimed to improve the maintenance plan for a Baumer HI VAC MX II B0112 autoclave used for sterilization in a hospital setting, classified as a Category V pressure vessel under NR 13. The methodology included an analysis of the maintenance history (2022–2023), identification of critical components, implementation of predictive maintenance techniques (non-destructive testing and visual inspection), optimization of the maintenance schedule, improvement in inventory management, and the implementation of a data recording and analysis system. The results showed an increase in maintenance performance indicators, with a 23.4% improvement in MTBF (Mean Time Between Failures), rising from 255.5 to 315.22 hours. Additionally, there was a 29% reduction in MTTR (Mean Time to Repair), decreasing from 18.65 to 13.25 hours, and an increase in equipment availability from 92.37% to 95.97%. Improvements included predictive maintenance, schedule optimization based on actual wear, integrated inventory management, and compliance with regulations such as RDC 15. Despite the short post-improvement analysis period (6 months), the project demonstrated potential to enhance operational efficiency, reduce costs, and improve safety. Continuous monitoring, full implementation of the proposed actions, and expansion of the predictive maintenance plan are recommended. The project contributed to the optimization of sterilization processes, positively impacting the quality of patient care and the safety of medical procedures.
Aplicação dos metamateriais acústicos na transmissão sonora
(2024-12-18) Silva, Bruno Cordeiro e; Fagundes Neto, Marlipe Garcia; Fagundes Neto, Marlipe Garcia; Kitatani Junior, Sigeo; Rodrigues, Marcos Vinicius Silva
Metamaterials are structures designed to exhibit physical properties not found in natural materials, such as controlled manipulation of sound waves, enhancing acoustic mitigation. When combined with additive manufacturing, this technology offers new possibilities for acoustic insulation. This study aims to explore the concept and application of the acoustic metamaterial model UOM (Ultra-open Metamaterial) proposed by Ghaffarivardavagh (2018), validating its efficiency in sound transmission loss based on its geometry. Additionally, the research investigates the advantages of simulations applied to acoustic silencers, evaluating their accuracy through impedance tube tests in conjunction with additive manufacturing of the silencer. The results indicate that the study's objectives were achieved, showing alignment between analytical, simulated, and tested results, with error margins ranging from 0.4% to 8.0%. The tests also showed proximity to the simulations, with a 3.5% error in the frequencies of maximum transmission loss, despite differences in peak absolute values. However, the fabrication of physical models using 3D resin printing revealed structural weaknesses due to reduced dimensions, highlighting challenges to be addressed for improved practical feasibility.
Deep learning para reconhecimento de sinais da LIBRAS como tecnologia assistiva
(Universidade Federal de Goiás, 2024-12-12) Pedrosa, Samuel França da Costa; Santana, Adriano César; Santana, Adriano César; Castro, Marcelo Stehling de; Nerys, José Wilson Lima
Communication is essential for social inclusion, yet the lack of linguistic accessibility often marginalizes specific groups, such as the Brazilian deaf community. This project proposes the use of deep learning to recognize signals from Brazilian Sign Language (LIBRAS) with the aim of translating them, through model predictions, into written Portuguese. Initially, videos of people signing were collected to form a dataset, which was subjected to processes of extraction and mapping of key body points using open-source tools provided by MediaPipe. The extracted data was processed and used as input for two designed models: one based on Long Short-Term Memory (LSTM) and another on Transformers. The study revealed that model performance is influenced by the alignment methods applied during data processing. The Transformer demonstrated superior results in terms of accuracy and generalization, albeit with higher computational demands. Conversely, the LSTM model showed satisfactory performance in terms of computational efficiency but exhibited limitations as classification complexity increased. One of the primary challenges was the difficulty in building a rich and robust dataset, due to the scarcity of available content for collection and extraction, especially when compared to other natural languages, whether textual or spoken. This limitation partially restricted the models' generalization capabilities. Despite these challenges, the project achieved promising results, suggesting that with enhanced and expanded datasets, its application as assistive technology can be extended to more complex scenarios with broader applicability. This study represents an advancement in the use of deep learning to promote inclusion and accessibility for the Brazilian deaf community.
Representação simplificada de um alimentador de média tensão da região metropolitana de Goiânia
(Universidade Federal de Goiás, 2024-12-12) Nascimento, Guilherme Pereira do; Lima, Ian Douglas Jacob de; Kopcak, Igor; Kopcak, Igor; Brigatto, Gelson Antônio Andrêa; Garces Negrete, Lina Paola
Studies on electric power systems, given its physical dimensions and power ratings, as well as its economic and social relevance, are typically conducted through computational simulations based on mathematical models of electrical grids, generators, loads, substations, and other components. The limitations regarding “bench testing” and even field testing create a demand for test systems that represent real networks for computational simulations, which can support various types of analysis. In this regard, benchmark test systems have historically been used as reference models. Depending on the objective or phenomenon to be analyzed, there are specific reduced test systems whose mathematical models are more appropriate for focusing on studies related to operational planning, system stability, and network protection security, among other possibilities. This paper develops a reduced equivalent of an electrical grid that serves the metropolitan area of Goiânia, based on data available in the literature as well as data requested from the local utility company, and it is representative of the dynamic behavior of this small portion of the National Interconnected System (SIN).
Avaliação da sobreamostragem de dados de voz na classificação automática da doença de Parkinson
(Universidade Federal de Goiás, 2024-12-19) Silva, Matheus Isac da; Felix, Juliana Paula; Felix, Juliana Paula; Silva, Karina Rocha Gomes da; Salvini, Rogerio Lopes
This study investigates a possible bias in oversampling via data windowing of vocal signals. Previous studies indicate that there is a bias for gait data when the data is treated independently, in addition there are statistical studies that show that data from the same individual carry similar information. An approach based on three databases containing vocal signals was used, two of which were unbalanced and one balanced. The K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Linear Discriminant Analysis (LDA), Naive Bayes and Decision Tree (DT) algorithms were applied, with pre-processing using StandardScaler and PCA behavior analysis. Cross validation was done with k-fold Cross Validation, with k=5, in all 3 bases, adapted for scenarios with and without bias in the training data. Models evaluated without considering bias showed inflated performances, while the rigorous approach showed more modest results. It is concluded that samples from the same individual in training and testing can inflate the performance of models, and it is crucial to apply oversampling correctly to develop reliable models for diagnosing PD.

Navegar

Submissões Recentes