Navegando por Data de Publicação, começando com "2026-04-23"

Filtrar resultados por ano ou mês

Agora exibindo 1 - 7 de 7

Dinâmica populacional do psilídeo e incidência de hlb em laranja valência conduzida em diferentes porta-enxertos, espaçamentos e estratégias de controle do vetor
(Universidade Federal de São Carlos, 2026-04-23) Cavalcanti, José Fabio Macêdo; Cristofani-Yaly, Mariângela; https://lattes.cnpq.br/7829946393687571; https://orcid.org/0000-0003-0283-989X; https://lattes.cnpq.br/6382519991127809; https://orcid.org/0009-0008-2208-9401; Schinor, Evandro Henrique; Azevedo, Fernando Alves de; https://lattes.cnpq.br/2162760988837984; https://lattes.cnpq.br/0574692206149746; https://orcid.org/0000-0002-9237-1460; https://orcid.org/0000-0003-3760-5030
Huanglongbing (HLB) is the most restrictive disease in citrus farming worldwide, requiring rigorous management of its insect vector, the psyllid Diaphorina citri. This study aimed to evaluate integrated management strategies for the vector and HLB in a young Valencia orange orchard under the influence of different rootstocks (citrandarins IAC 3155, Barretos, Itajobi, Santamélia and the citrumelo Swingle), planting spacings (1.5, 2.0, 2.5 and 3.0 m) and control strategies (physical barrier with processed kaolin and chemical control with thiamethoxam via drench). The experiment was conducted at the Sylvio Moreira Citrus Center in Cordeirópolis, SP, evaluating vegetative development, sprouting dynamics, psyllid population fluctuation, HLB incidence and severity, as well as the residual effect of the insecticide. The results demonstrated that vigorous rootstocks (IAC 3155 and IAC 3073 Barretos) induced greater canopy growth, while IAC 3026 Santamélia confirmed its dwarfing characteristic. Higher planting density (1.5 m between plants in the row) accelerated coverage in the planting row and reduced the total intensity of sprouting. A strong positive correlation was found between sprouting peaks and population outbreaks of D. citri, with the Swingle and IAC 3073 Barretos rootstocks favoring greater attraction of the vector. Fortnightly spraying with kaolin acted as an effective physical and visual barrier, reducing the insect population density by up to 72% during peaks. Although the incidence of HLB remained high (>90%) throughout the area, kaolin significantly reduced disease severity and symptom progression, with optimized protection in denser plantings. In chemical control, the rootstock variety did not influence the insecticide's efficiency, which showed excellent control and residual effect for 56 days, covering two reinfestation cycles. From the 57th day onwards, the depletion of the molecule's systemic protection in the plant reduced the product's effectiveness. It is concluded that the adoption of less vigorous rootstocks, combined with denser planting and the combination of kaolin with systemic insecticides, constitutes a strategy for mitigating the HLB-psyllid complex.
Mecanismos de trincamento por fadiga em revestimentos asfálticos reforçados com geogrelhas: interação entre aderência de interface e rigidez do reforço
(Universidade Federal de São Carlos, 2026-04-23) Santos, Karolina Maria dos; Silva, Matheus Pena da Silva e; https://lattes.cnpq.br/8819593174579282; https://orcid.org/0000-0002-7009-4458; Correia, Natália de Souza; https://lattes.cnpq.br/2172645507238077; https://orcid.org/0000-0002-6867-9596; https://lattes.cnpq.br/2462457683227987; https://orcid.org/0009-0000-7055-4496; Freire, Reuber Arrais; Almeida, Mario Sergio de Souza; https://lattes.cnpq.br/2916580036912769; https://lattes.cnpq.br/4157408069497314; https://orcid.org/0000-0001-9056-0463; https://orcid.org/0000-0002-0222-4804
The incorporation of geosynthetic reinforcements at the interface between asphalt layers has been widely adopted as a strategy to extend pavement service life, contributing to the reduction of permanent deformation, the improvement of fatigue resistance, and the mitigation of reflective crack propagation. In this context, numerous studies have been conducted to better understand the mechanical behavior of reinforced asphalt systems and to provide guidance for the appropriate selection and design of these materials. In particular, the literature has extensively discussed the effects of geosynthetic reinforcement inclusion on flexural performance and interface bonding conditions between asphalt layers. However, significant knowledge gaps still remain regarding the mechanisms governing the effectiveness of these reinforcements throughout the pavement service life. In particular, there is a need for a deeper understanding of the interaction between interface bonding and reinforcement stiffness, as well as the influence of these parameters on the initiation, evolution, and propagation of damage within reinforced asphalt systems. This limitation hinders the advancement of mechanistic approaches and the development of rational design methods for geogrid-reinforced asphalt pavements. In this context, the present study investigates the combined influence of interface bonding and geogrid stiffness on the fatigue performance of reinforced asphalt systems at different stages of their service life, using Digital Image Correlation (DIC) as a strain-fieldbased evaluation tool. For this purpose, an 80 m experimental section was constructed on SP-225 State Highway, comprising one unreinforced segment and three reinforced sections using polyester (G-PET), polyvinyl alcohol (G-PVA), and fiberglass (G-FV) geogrids. Specimens extracted from the field section were tested in the laboratory through Leutner-type interlayer shear tests and four-point bending fatigue tests (4PBT) under cyclic loading with DIC monitoring. The present study has advanced the understanding of the mechanisms governing the fatigue of asphalt pavements reinforced with geogrids, demonstrating that structural performance results from the interaction between interlayer bonding, reinforcement properties, and loading levels. The results highlight that the analysis of these systems under cyclic loading requires an integrated approach, combining traditional fatigue parameters, improvement factors, and image-based analysis via DIC. This integration was essential to explain results that might appear contradictory when evaluated in isolation. Four primary crack reflection mechanisms were identified. The results indicated that the effectiveness of the reinforced systems depends on the loading level, with improved performance observed at the lower load levels tested, which are more representative of actual pavement service conditions. The effectiveness of the reinforced systems was found to be strongly dependent on maintaining adequate levels of interlayer bonding, which is essential for the mobilization of the reinforcement and the transfer of stresses throughout the loading process. The study concluded that interface bonding is the key parameter governing reinforcement mobilization and the resulting performance gains, controlling the efficiency of crack mitigation systems regardless of the stiffness of the reinforcement employed. Overall, the performance of these systems cannot be attributed to a single isolated parameter, but rather to the complex interaction among geosynthetic stiffness, interface shear strength, applied stress level, and the evolution of damage mechanisms throughout the service life of the pavement structure.
As diferenças nos processos de tomada de decisão da inteligência artificial e dos seres humanos no laboratório jogo de xadrez
(Universidade Federal de São Carlos, 2026-04-23) Andrade, Leo Pasqualini de; Valejo, Alan Demétrius Baria; https://lattes.cnpq.br/9546164790189830; https://orcid.org/0000-0002-9046-9499; Iasulaitis, Sylvia; https://lattes.cnpq.br/4226275371443693; https://orcid.org/0000-0002-3526-1003; https://lattes.cnpq.br/0142306249947881; https://orcid.org/0000-0002-8893-0338; Gracioso, Luciana de Souza; Amaral, Roniberto Morato do; Pereira, Eanes Torres; Brusamolin, Valério; Amorim, Cláudio Alves de; https://lattes.cnpq.br/4898201916360294; https://lattes.cnpq.br/6958372164719600; https://lattes.cnpq.br/2030738304003254; https://lattes.cnpq.br/2076541861375113; https://lattes.cnpq.br/4976012823452609; https://orcid.org/0000-0002-6320-4946; https://orcid.org/0000-0002-9816-231X; https://orcid.org/0000-0002-3849-0573; https://orcid.org/0000-0002-9717-794X
Artificial Intelligence is an interdisciplinary field of study. Pioneers in computer science, mathematicians, and engineers envisioned the possibility of a computer simulating human thought through different computational methods, for example, artificial neural networks, inspired by biological research in the then-embryonic neuroscience. Philosophers and psychologists joined the field's research, addressing questions about the possibility of artificial intelligence thinking and having consciousness. The testing ground for artificial intelligence was chess, an ancient game with simple rules and complete information, which, in order to play it, reveals elements of human reasoning, such as selective attention, pattern recognition, abstractions, and logical reasoning. The calculations of variations in chess are performed with great accuracy by current artificial intelligence algorithms such as Stockfish and AlphaZero, but still present questions to be resolved by computer scientists in order to reach solutions conceived by human abstraction. Although the performance of chess engines has been extensively studied from a computational perspective, there is a gap in the literature regarding the systematic comparison between human and algorithmic decision-making processes in the same experimental environment. The objective of this thesis is to understand the differences, symmetries, and controversies between artificial intelligence and human intelligence in decision-making, based on a chess game laboratory. This thesis contributes to the literature by offering a novel interdisciplinary comparative analysis of human and algorithmic decision-making processes in the chess game laboratory, articulating empirical evidence, cognitive foundations, and sociotechnical interpretation. The methodological steps used were based on Bruno Latour's Actor-Network Theory, which proposes the idea of actants, where humans and non-humans join together in a network of actions. A literature review was conducted using the keywords chess, intelligence, and decision-making; an online survey was carried out with chess players on aspects of pattern recognition, abstraction, and to obtain opinions on problem situations in the context of chess. Semi-structured interviews were also conducted with computer scientists, neuroscientists, and chess masters to discuss the topic of artificial intelligence and the game of chess. The results indicated that, while chess engines predominantly compute using probabilistic calculations and tree search, humans mobilize pattern recognition from their long-term memory acquired through study and learning, as well as emotional and contextual components. It was observed that chess players with greater playing strength showed a greater ability to recognize patterns, while less experienced players resorted to mental calculations. The interviews revealed controversies among computer scientists, neuroscientists, and chess masters regarding the notion of engines simulating emotions and making abstractions. Conclusion: although chess algorithms have already surpassed humans in the game, controversies are still observed regarding these programs' understanding of game situations and their failure to consider artistic-cultural, ethical, and emotional biases—characteristics associated with human behavior and creativity. On the other hand, chess programs are tools that assist masters in their analyses before and after games and help players of all levels as a source of leisure and learning.
Raw data used for determining the ionization energies of dimers of small molecular clusters
(Universidade Federal de São Carlos, 2026-04-23) Souza, Gabriel Luiz Cruz de; Peterson, Kirk Arnold; Santos, José Luiz Felix; https://lattes.cnpq.br/2821352329397835; https://lattes.cnpq.br/3442700129094678; https://lattes.cnpq.br/2951756857186266; https://orcid.org/0000-0002-2192-7345; https://orcid.org/0000-0003-4901-3235; https://orcid.org/0000-0003-1562-5233
Modelagem matemática da reforma a vapor do etanol
(Universidade Federal de São Carlos, 2026-04-23) Sgobi, Marcus Vinicius; Ribeiro, Marcelo Perencin de Arruda; https://lattes.cnpq.br/0381402687491195; https://lattes.cnpq.br/5091119086062042; Cruz, Antonio José Gonçalves da; Nucci, Edson Romano; https://lattes.cnpq.br/1812806190521028; https://lattes.cnpq.br/4274582376583918
The historical global dependence on fossil fuels and the growing concern regarding the worsening of the greenhouse effect demand the diversification of the energy matrix towards more sustainable alternatives. In this scenario, hydrogen stands out as a promising energy vector due to its high energy density and the absence of greenhouse gas emissions during its use. Although current hydrogen production is dominated by fossil methane steam reforming, steam reforming of ethanol emerges as a new sustainable route. To make steam reforming of ethanol viable, supported catalysts, such as Nickel and Cobalt-based bimetallics, have shown great potential in inhibiting coke formation and increasing the H2 yield. However, the high chemical complexity of the ethanol molecule generates various parallel and consecutive reactions, which hinders the elucidation of the process kinetics. Based on this context, the present work proposed a reaction pathway and, from it, developed a phenomenological kinetic model capable of describing steam reforming of ethanol for hydrogen production. Using experimental screening data previously obtained by the research group, the study implemented and compared three different algorithms for the parametric fitting of the developed model: Non-Linear Least Squares, Simulated Annealing, and the Covariance Matrix Adaptation Evolution Strategy. Before the parametric fitting, a thermodynamic analysis proved that the experimental screening tests operated far from equilibrium. The fitting results highlighted the limitations of traditional deterministic optimizations, emphasizing the robustness of the evolutive algorithm in bypassing local minima and performing uncertainty estimation through confidence intervals. Although the preliminary data were limited, they proved to be sufficient to estimate the initial kinetic parameters. These estimates were for the 4Co4Ni, 8Co,
Comportamento estrutural de pilares de aço preenchidos com concreto de agregado reciclado cerâmico submetidos à compressão axial
(Universidade Federal de São Carlos, 2026-04-23) Prates, Lucas Mognon Santiago; Ferreira, Felipe Piana Vendramell; https://lattes.cnpq.br/5933034803347448; https://orcid.org/0000-0001-8007-789X; Martins, Carlos Humberto; https://lattes.cnpq.br/1665294496138141; https://orcid.org/0000-0001-7342-5665; https://lattes.cnpq.br/7177250247199481; https://orcid.org/0009-0003-0955-1378; Lima, Luciano Rodrigues Ornelas de; Nicoletti, Renato Silva; Bolandim, Emerson Alexandro; Diógenes, Hidelbrando Jose Farkat; https://lattes.cnpq.br/2151919861612154; https://lattes.cnpq.br/3109388079611718; https://lattes.cnpq.br/9169305081866640; https://lattes.cnpq.br/6693686167483212; https://orcid.org/0000-0002-7332-3818; https://orcid.org/0000-0002-3021-0944; https://orcid.org/0000-0001-6220-7646; https://orcid.org/0000-0003-2480-7688
Concrete-filled steel tubular columns are an efficient structural solution for compression members, since the interaction between the steel tube and the concrete core contributes to increased strength, stiffness, and ductility. At the same time, the growing demand for more sustainable solutions in the construction sector has encouraged the use of industrial waste as a replacement for natural aggregates. In this context, the feasibility of using recycled ceramic sanitary ware aggregate as a replacement for natural sand in self-compacting concrete for steel composite columns subjected to axial compression is assessed, considering its effects on material properties, structural behavior, and strength prediction by numerical and design code models. For this purpose, the constituent materials were characterized, and two self-compacting concrete mixes were analyzed, with natural fine aggregate replacement levels ranging from 0 to 100% in the first mix and from 0 to 50% in the second, in 25% increments. Subsequently, 24 experimental axial compression tests were carried out on circular concrete-filled steel tubular columns, comprising 16 short columns and 8 columns of intermediate slenderness. In addition, a finite element numerical model was developed in ABAQUS, validated against the experimental results, and used in a parametric study with 1,442 models, varying geometric and mechanical parameters as well as the recycled aggregate replacement level. The results indicate that the incorporation of recycled aggregate reduces the workability of the mixtures, without preventing the production of self-compacting concretes with satisfactory performance. It was found that increasing the replacement level does not necessarily lead to strength loss, with strength gains observed at certain replacement levels, depending on the mix composition. In the composite columns, the structural behavior was consistent with steel-concrete composite action, with efficient filling of the tube and mobilization of confinement in the short columns, whereas, in the slender members, the response was predominantly governed by global instability. Among the replacement levels evaluated, the 25% replacement level showed the most balanced performance for structural application, as it maintained a more stable response under different conditions. The numerical modeling satisfactorily reproduced the experimental response, and the parametric study confirmed global slenderness as the variable with the greatest influence on member efficiency. It is concluded that the use of recycled sanitary ware ceramic aggregate in self-compacting concrete for filling circular concrete-filled steel tubular columns is technically feasible and structurally promising, combining satisfactory mechanical performance, reuse of industrial waste, and potential application in structural concretes.
Detecção de objetos de trânsito da simulação para o mundo real: um estudo de adaptação de domínio não supervisionada em nível de pixel e de características
(Universidade Federal de São Carlos, 2026-04-23) Medeiros, André Neves; Almeida, Tiago Agostinho de; http://lattes.cnpq.br/5368680512020633; http://lattes.cnpq.br/6707942601520564; Papa, João Paulo; Valejo, Alan Demétrius Baria; Backes, André Ricardo; Lorena, Ana Carolina; Naldi, Murilo Coelho; Almeida, Tiago Agostinho de; http://lattes.cnpq.br/9039182932747194; http://lattes.cnpq.br/9546164790189830; http://lattes.cnpq.br/8590140337571249; http://lattes.cnpq.br/3451628262694747; http://lattes.cnpq.br/0573662728816861; http://lattes.cnpq.br/5368680512020633
Training object detection models for real-world applications often requires large amounts of annotated data, which can be costly and time-consuming to obtain. Synthetic datasets provide a scalable alternative, enabling automatic annotation and controlled data generation. However, models trained exclusively on synthetic data typically suffer from performance degradation when applied to real-world images due to domain shift. This dissertation investigates simulation-to-real domain adaptation for object detection, focusing on the combination of pixel-level and feature-level adaptation strategies. At the pixel level, CycleGAN is used to translate synthetic images into the visual style of real-world datasets, reducing low-level discrepancies between domains. At the feature level, the ConfMix framework is adopted as a baseline for unsupervised domain adaptation, leveraging confidence-based mixing between labeled source samples and pseudo-labeled target samples. Building upon these approaches, this work proposes TransConfMix and TransConfMix-2Stage variants, which incorporate translated synthetic images into the adaptation process, and a KL-regularized extension (ConfMix-Distill) that enforces consistency between predictions obtained from original and translated versions of the same sample. These modifications aim to improve adaptation stability and enhance model generalization under domain shift. Experiments are conducted on multiple synthetic-to-real scenarios, with DOLPHINS and SIM10K as source domains and Cityscapes and nuScenes as target domains. The results demonstrate that feature-level adaptation consistently improves over the source-only baseline across the evaluated scenarios, and that pixel-level translation can further support this process when translation quality is sufficiently high. However, the proposed hybrid extensions did not consistently surpass the original ConfMix framework, suggesting that the interaction between pixel- and feature-level adaptation is highly sensitive to domain characteristics and translation fidelity. Overall, this work provides a systematic analysis of the interaction between pixel-level and feature-level domain adaptation, showing that their combined effect in object detection is strongly dependent on translation fidelity and domain characteristics.