PPGEE https://repositorio.ufpb.br/jspui/handle/tede/4457 2026-10-12T03:30:06Z 2026-10-12T03:30:06Z Froes, Filipe Bulhões https://repositorio.ufpb.br/jspui/handle/123456789/39397 2026-10-03T06:10:26Z 2026-01-30T00:00:00Z

Título: Análise e desenvolvimento de modelos de machine learning para simulação de uma usina termelétrica de ciclo combinado Autor(es): Froes, Filipe Bulhões Orientador: Macêdo, Euler Cássio Tavares de Abstract: Artificial intelligence (AI) has proven to be a powerful tool for addressing the challenges of energy transition and the search for greater efficiency in power generation systems. In the Brazilian electricity sector, marked by the growing participation of natural gas-fired thermoelectric power plants, machine learning (ML) techniques can help optimize the operation of these plants, contributing to greater efficiency and reduced emissions. In this context, digital twin systems (DTS) emerge as a promising technology for simulation and decision support. The objective of this study was to develop and compare different machine learning models applied to the simulation of a combined cycle thermal power plant, in preparation for its integration into a DTS. Data processing, training, evaluation, and model comparison methodologies were explored, including individual algorithms and ensemble techniques such as bagging, boosting, stacking, and blending. The results obtained demonstrated that regression models such as SVR and GPR, as well as their variants in ensemble techniques, achieved better performance in terms of normalized mean absolute error and coefficient of determination. In addition, it was found that combinations of models can provide greater robustness and accuracy, indicating that hybrid approaches are promising alternatives for representing the behavior of the plant under different conditions. Editor: Universidade Federal da Paraíba Tipo: Dissertação

2026-01-30T00:00:00Z Aires, Virgínia Vieira https://repositorio.ufpb.br/jspui/handle/123456789/39391 2026-10-03T06:10:10Z 2026-02-27T00:00:00Z

Título: Study and implementation of a SmartTree in the IoT context aiming at real-time meteorological data measurement Autor(es): Aires, Virgínia Vieira Orientador: Souza, Cleonilson Protasio de Abstract: Given the growing need for innovative and accessible environmental monitoring solutions, especially in hard-to-reach areas, this study investigates the use of temperature sensors installed on a structure inspired by natural trees, referred to as the SmartTree, located at the Center for Alternative and Renewable Energies (CEAR) of the Federal University of Paraíba, Brazil, to estimate meteorological parameters such as wind speed and direction. Motivated by this context, the study proposes an alternative approach to conventional meteorological monitoring by employing tree inspired structures as natural platforms for climate sensing, based on the hypothesis that internal temperature variations within the tree trunk, captured by sensors installed at different heights, depths, and orientations, indirectly reflect external patterns of wind and ambient temperature. The methodology is based on the application of the Discrete Wavelet Transform to extract multiscale features, which are used as input for machine learning models, including Random Forest, Gradient Boosting, Support Vector Machine, and Linear Regression. Data were collected and complemented by additional information from a nearby weather station. Among the evaluated models, Random Forest achieved the best performance, reaching high accuracy with a mean absolute percentage error of 6.60% for wind direction. The results confirm the potential of the tree inspired solution as a viable alternative for climate monitoring, with important applications such as forest fire prevention. Editor: Universidade Federal da Paraíba Tipo: Dissertação

2026-02-27T00:00:00Z Matos, Vinício de Araújo https://repositorio.ufpb.br/jspui/handle/123456789/38655 2026-08-12T06:12:19Z 2025-12-08T00:00:00Z

Título: Análise e modelagem das emissões instantâneas de co2 em veículos de combustão interna por meio de dados OBD-II e redes neurais artificiais Autor(es): Matos, Vinício de Araújo Orientador: Villanueva, Juan Moises Maurício Abstract: This study develops a methodology to evaluate and model vehicle carbon dioxide (CO2) emissions using data collected via the OBD-II interface, integrating detailed automotive analysis with artificial neural network techniques. The overall goal is to compare the emission behavior of internal combustion engine vehicles with electric vehicles under real operating conditions. Relevant operational data (speed, throttle position, air flow, etc.) were gathered directly from vehicles through OBD-II and then explored to identify patterns and correlations (such as the strong relationship between engine load and CO2 generation). Predictive models using Multi-Layer Perceptron (MLP) and Long Short-Term Memory (LSTM) neural networks were trained to estimate instantaneous CO2 emissions. The LSTM model achieved the best performance with a mean absolute percentage error (MAPE) of around 5%, significantly outperforming the MLP (MAPE 16%). The comparative analysis indicates that electric vehicles eliminate direct emissions during use, but their total CO2 impact depends on the electrical grid mix and manufacturing emissions. In conclusion, the proposed approach - combining OBD-II data and neural networks - proved effective for real-time emission estimation and can complement standardized lab-based estimates. Editor: Universidade Federal da Paraíba Tipo: Dissertação

2025-12-08T00:00:00Z Gonçalves, Luiz Felipe Rodrigues https://repositorio.ufpb.br/jspui/handle/123456789/38049 2026-05-14T06:07:43Z 2025-12-15T00:00:00Z

Título: Desenvolvimento de um sensor virtual de vazão utilizando redes neurais para macromedição de uma estação de distribuição de água Autor(es): Gonçalves, Luiz Felipe Rodrigues Orientador: Villanueva, Juan Moises Maurício Abstract: Law No. 14,026/2020, which established the new legal framework for sanitation in Brazil, introduced significant changes for water and sewage distribution companies by imposing stricter regulations and increasing competition in the sector. One of its key objectives is to reduce water losses and improve macromeasurement rates. According to the National Sanitation Information System (SNIS), the average water loss in Brazil is around 40%, while only 70% of the distributed water is properly measured. However, the high cost of sensors, especially flow macrometers, limits their widespread deployment, leading many companies to focus monitoring efforts on metropolitan areas. To address this issue, indirect measurement methods allow flow estimation based on network pressure and reservoir level data, reducing the need for physical sensors across the entire distribution system. In this context, artificial intelligence (AI) has been increasingly applied to improve the efficiency of water supply systems, enabling the reconstruction of missing data, fault detection, and forecasting of operational variables. Soft sensors based on neural networks allow continuous system monitoring and anomaly detection, enhancing operational decision-making and minimizing resource waste. This study presents the development and implementation of a soft sensor based on Long Short-Term Memory (LSTM) neural networks to estimate flow rates in a water supply system. The model was trained using flow, pressure, and reservoir level data collected by the Companhia de Água e Esgotos da Paraíba (CAGEPA) in the municipality of Salgado de São Félix, through an MQTT broker and cloud-based monitoring. The model validation was performed by comparing the estimated values with real measurements from physical sensors, resulting in a mean absolute percentage error lower than 5%, demonstrating the reliability of the proposed approach. Editor: Universidade Federal da Paraíba Tipo: Dissertação

2025-12-15T00:00:00Z