Repositório UFSCar

O Repositório Institucional da UFSCar (RI UFSCar) é um sistema de informação que visa armazenar, preservar, organizar e disseminar amplamente a produção intelectual dos diversos setores e segmentos da comunidade da UFSCar, provendo o acesso aberto à informação produzida na instituição e registrada como científica, tecnológica, didática, artístico-cultural e técnico-administrativa.

Para mais informações, acesse: RI UFSCar

Instruções aos autores: Clique aqui.

Comunidades

Selecione uma comunidade para navegar por suas coleções

Agora exibindo 1 - 5 de 5

Submissões Recentes

Estudo comparativo da eficácia da boretação sólida com a boretação a plasma na liga Ti6Al4V
(Universidade Federal de São Carlos, 2022-08-31) Silva, Felipe Lopes Fonseca da; Rossino, Luciana Sgarbi; http://lattes.cnpq.br/0139027055418391; https://lattes.cnpq.br/8545503787499646; Correa, Diego Rafael Nespeque; Rangel, Rita de Cássia Cipriano; Rossino, Luciana Sgarbi; http://lattes.cnpq.br/2282008544774759; http://lattes.cnpq.br/9805077380369331; http://lattes.cnpq.br/0139027055418391
Ti6Al4V alloy is one of the most widely used alloys in the biomedical field as an implant and in the aerospace field for structures. However, surface treatment of these alloys is extremely important, since in the biological field Ti6Al4V alloy contains aluminum and vanadium, which are toxic to the biological environment, while in industrial areas the surface properties of the material influence the performance of the structure. Boriding is one of the surface treatments that has been attracting the attention of researchers due to the properties acquired in materials after treatment, and the solid pack boriding and plasma paste boriding processes are the most widely used in economic and toxicological terms. This work aims to perform a comparative analysis of the surface properties acquired by the conventional pack boriding method and the plasma paste boriding method, which was developed in this work, on the Ti6Al4V alloy. In the first stage of this work, Ti6Al4V samples were treated by pack solid boriding at 750, 850, 950, and 1050ºC for 3 and 6 hours using the boriding agent Ekabor 2. Paste plasma boriding treatment was carried out in a vacuum reactor using a gas ratio of 40% H2, 40% N2 and 20% Ar with a boriding agent supplied by a solid paste rich in this element. The second stage involved the development of a plasma boriding treatment with solid paste, which included the study of four different pastes: paste 1 composed of 80% Ekabor 2® – 20% Na2B4O7, paste 2 with 100% Na2B4O7, paste 3 with 70% Na2B4O7 – 30% SiC, and paste 4 with 30% Na2B4O7 – 70% SiC. After being molded onto the samples, these pastes were subjected to a plasma treatment at 650°C for 3 hours. For the third stage of this work, the plasma boriding treatment was carried out at 650°C and 700°C for 3 and 6 hours respectively, using the paste that demonstrated the best results in terms of stability and characteristics of the formed layer. In the pack solid boriding process, boron diffusion was identified at temperatures of 850, 950, and 1050ºC in 3 and 6 hours, with the formation of a layer containing the TiB2 (titanium diboride) and TiB (titanium boride) phases. However, in all samples treated in the pack solid boriding process, the presence of pores and the formation of the TiO2 phase were observed. This oxide is formed due to the high affinity of oxygen for titanium, and this phase hinders the regular diffusion of boron in the substrate and weakens the layer. At 950ºC for 3 and 6 hours, the layers showed better regularity, thicknesses of 89.052 ± 5.462 µm and 76.107 ± 4.262 µm, surface hardness of 1247.9 ± 120.9 HV and 1025.4 ± 46.3 HV respectively, and the formation of TiB2 and TiB phases. In plasma boriding, the pastes with the highest concentrations of borax (Na2B4O7), pastes 2, 3, and 4, produced a layer without pores, with the formation of TiB2 and TiB phases, without the presence of the TiO2 phase. Paste 3 showed the best results, with a layer thickness of 11.602 ± 0.436 µm, a hardness of 663.9 ± 86.6 HV, and the presence of TiB2 and TiB phases. The condition of paste 3 was used for the development of the third stage of this work. It was observed that in terms of layer thickness, hardness, boron concentration, and phases formed, the treatments performed for 3 hours presented the best results at 650 and 700ºC, with a total layer thickness of 11.602 ± 0.436 µm and 6.429 ± 0.619 µm, respectively, with hardnesses of 663.9 ± 86.6 HV and 768.3 ± 50.7 HV, and TiB2 and TiB phases formed. The plasma boriding process showed better results compared to the solid process due to it managed to diffuse the boron, forming the desired layer with the TiB2 and TiB phases present, without defects or porosity, and without the formation of the TiO2 phase.
Um estudo sobre diferentes métodos de classificação para dados extremamente desbalanceados sob a ótica do aprendizado sensível ao custo
(Universidade Federal de São Carlos, 2025-02-25) Angélico, João Pedro Modesto; Diniz, Carlos Alberto Ribeiro; https://lattes.cnpq.br/3277371897783194; https://orcid.org/0000-0003-3464-1108
In classification tasks, we often encounter datasets with highly imbalanced classes, such as those found in rare disease diagnosis, financial fraud detection, and industrial failure analysis. Training traditional models on such datasets tends to favor the majority class, resulting in poor ability to identify observations from the minority class, which is usually the one of greatest interest. In this undergraduate thesis, we present and compare different classification models — Logistic Regression, Random Forest, and XGBoost — from the perspective of cost‑sensitive learning, applying class weights to compensate for severe imbalance. Using a real‑world credit card transaction dataset (with only 0.172% fraud cases), we demonstrate how assigning differential weights to classification errors significantly improves the sensitivity (recall) in detecting the minority class. Furthermore, we evaluate which combination of model and validation technique (data‑split or stratified cross‑validation) yields the best trade‑off between sensitivity, precision, and F1‑score. The results indicate that XGBoost with appropriate weighting achieves the best overall performance, reaching recall above 0.89 and F1‑score close to 0.88.
Análise comparativa do desempenho de modelos de machine learning na previsão de focos de incêndio no cerrado utilizando variáveis climáticas
(Universidade Federal de São Carlos, 2026-04-01) Pádua, Juliano Eleno Silva; Levada, Alexandre Luis Magalhães; http://lattes.cnpq.br/3341441596395463
Predicting fire outbreaks in the Cerrado biome using meteorological variables is a relevant problem for environmental monitoring and supporting alert systems. In this work, a comparative analysis of classical machine learning models for hourly prediction of fire outbreak occurrence was carried out, based on the integration between INMET data and records from INPE's BDQueimadas. The Logistic Regression, Naive Bayes, linear SVM, Random Forest, and XGBoost models were evaluated in different data preparation scenarios, including original datasets, datasets with derived variables, KNN imputation, and imbalance treatment strategies such as SMOTE and weight balancing. The results showed that tree ensemble-based models were the most suitable for the problem, especially XGBoost, and that feature engineering and explicit imbalance treatment contributed decisively to increased performance, especially in metrics more sensitive to the detection of the positive class, such as PR - AUC and F1-score.
CPI da pirataria: economias populares, disputas e representações sobre o mercado de falsificações
(Universidade Federal de São Carlos, 2025-07-09) Servilha Filho, Tiago Tadeu; Rangel, Felipe; http://lattes.cnpq.br/1619003785230081; https://orcid.org/0000-0002-0679-3756; http://lattes.cnpq.br/3883896580869658
This research aimed to understand, through the analysis of the Parliamentary Inquiry Committee (CPI) on Piracy conducted by the São Paulo City Council, the representations, disputes, and discourses surrounding the counterfeit goods market. By doing so, it advances the understanding of the tensions around divergent perceptions of the counterfeit market, enabling an analysis of the affinities between state institutions and large corporations in anti-piracy efforts, as well as their effects on the control of popular economic practices. Methodologically, the investigation was based on an in-depth analysis of the stenographic records, videos, and documents submitted to the CPI, as well as a survey of news regarding the production, trade, and anti-piracy actions. The research drew on existing Social Sciences literature on popular economies and the issue of piracy to analyze the discourses, representations, and emerging conflicts within the CPI, interpreting them within the framework of economic disputes surrounding contested markets.
Proposta e validação de uma taxonomia de variações ortográfcas em tweets do mercado financeiro
(Universidade Federal de São Carlos, 2025-07-25) Scandarolli, Clarissa Lenina; Di-Felippo, Ariani; http://lattes.cnpq.br/8648412103197455; https://orcid.org/0000-0002-4566-9352; http://lattes.cnpq.br/6804082969710510; https://orcid.org/0009-0009-9142-6771
Given the relevance of the Twitter platform (now X) for various segments of society, Natural Language Processing (NLP) tools and applications capable of handling the primarily non-canonical language of the tweet genre (currently referred to as posts) are in high demand. To develop them, annotated corpora (known as tweetbanks) are essential resources, as are the description and analysis of their linguistic characteristics. In this Undergraduate Thesis, the object of investigation was the DANTEStocks corpus of 4,048 financial market tweets, which is the first in Portuguese to be annotated according to the Universal Dependencies (UD) model. More precisely, a survey of orthographic phenomena in the corpus was conducted, treated as "variations" rather than "errors", in accordance with the theoretical assumptions of Variationist Sociolinguistics. These phenomena were systematized into a hierarchical typology with two dimensions: "Standard Norm" and "Innovative Norm", which seek to capture variations of canonical and innovative lexical forms. Based on this typology, the phenomena observed in 3,614 tokens (found in 1,069 tweets) from DANTEStocks were manually annotated, yielding a preliminary characterization of the corpus. The results evidenced the predominance of the Innovative Norm, accounting for 92.81% of the annotated phenomena (3,457 occurrences), reinforcing the hypothesis that, in User-Generated Content (UGC) on financial topics, innovative linguistic strategies prevail, manifesting in tokens that function as codes and systematic forms of digital communication, characteristic of the medium and of a particular social context. In this way, the annotation of graphic variations in DANTEStocks broadens the understanding of the language of financial market tweets and may enhance the tolerance of NLP models toward non-canonical language, enabling them to recognize variant forms as linguistically valid and semantically informative.