http://www.tede2.ufrpe.br:8080/tede2/handle/tede2/4344 Sat, 27 Sep 2025 12:06:01 GMT 2025-09-27T12:06:01Z http://www.tede2.ufrpe.br:8080/tede2/handle/tede2/8757 Título: Modelagem computacional de ecossistemas com competição por recursos e evolução em ambientes heterogêneos Autor: DAZA, Sara Lucia Castillo Primeiro orientador: OLIVEIRA, Viviane Moraes de Abstract: In this thesis we study an ecosystem model with a spatial structure in which species compete for resources and evolve in a heterogeneous environment. Environmental heterogeneity is introduced through the allocation of resources, which are distributed on the network through a fractal landscape generated through the simulation of the fractional Brownian movement. Thus, the roughness of the landscape is controlled by Hurst’s exponent, H. Each species is characterized by a set of half-saturation constants, which define the e ciency of the species in the use of each resource. Only one species is initially introduced into the system, and new species are generated from mutations that occur with probability ѵ. The set of half-saturation constants that characterize the mutant species is obtained from the set of the ancestral species, in which one of the constants is modified and obtained through a normal distribution in which the mean is equal to that of the ancestral species and three di erent values for the variance are studied. In the first part of the work, we studied the behavior of diversity patterns presented by the system. We observed that the mean diversity presented a lower value for H = 0:01 (very rough landscape) for the case in which the probability of mutation of the species is lower. We also verify that the species-area relationship has two power law regimes in which S ~ Az, where the exponents obtained for large areas are greater than those obtained for small areas. We also investigated the relationship between the mean number of species and the Hurst exponent, H. For the highest mutation probability value, we note that a higher value of variance in the distribution of the half-saturation constants leads to less diversity. For the case in which the probability of mutation is lower, we observed an increase in the average number of species with H, and less diversity for the case in which the variance of the distribution of the half-saturation constants is smaller. In the second part of the work we did a more statistical study, where we analyzed the behavior of the distribution of the fluctuations in the temporal evolution of diversity. We also studied the relationship between diversity and di erent mutation probability values. We saw that the stretched exponential distribution provided a good fit of the behavior of heavy tail distributions, as were the distributions of the histograms of increments of diversity. We find an adjustment exponent β ≈ 1 indicating that the system has memory for low mutation probabilities, and an exponent β = 2 for higher mutation probabilities, from which we infer that the system behaves like a Markov process. We also noticed a behavior change in the relationship between the exponent and the β mutation probability. For low mutation probability values this relationship follows a power law. For high mutation probabilities, β becomes independent of ѵ. We realized that changing the fluctuation in diversity only depends on the probability of mutation. Instituição: Universidade Federal Rural de Pernambuco Tipo do documento: Tese Thu, 20 Feb 2020 00:00:00 GMT http://www.tede2.ufrpe.br:8080/tede2/handle/tede2/8757 2020-02-20T00:00:00Z http://www.tede2.ufrpe.br:8080/tede2/handle/tede2/8756 Título: Das estatísticas espaciais, Krigagem e Gi* de Getis-Ord, na interpolação de dados e análise de Clusters em dados de qualidade de leite e raiva bovina Autor: FÉRRER, Moisés Tenório Primeiro orientador: MOREIRA, Guilherme Rocha Abstract: The objectives of this work were to analyze the spatio-temporal distribution of nervous diseases in cattle, identifying the presence of hot spot and cold spot of concentration of cases, to search for associations that explain the occurrence and to estimate the dynamics of the disease over time, in the period 2005-2018 and the spatial variability of the composition of chilled raw milk and to elaborate maps with interpolation of data of the physical-chemical specifications of the milk, in the state of Alagoas and in the mesoregion of Agreste Pernambucano, in 2014 and 2015 and Records of 30,929 were analyzed notifications of nervous diseases throughout Brazil included in the Continental Epidemiological Surveillance System (SivCont), from 2005 to 2018 and 3,863 official reports of refrigerated raw milk samples, collected from 432 tanks of direct expansion in the studied region. Regarding the species affected among the confirmed cases of rabies, the main one was bovine with 8,977 (85.43%) cases, followed by the equine species with 1,143 cases (10.87%). All other species had a relative frequency below 1%. Finally, it is concluded that there is spatial variability for fat, lactose, protein, total solids and defatted dry extract of chilled raw milk produced in the state of Alagoas and in the mesoregion of Agreste Pernambucano and that rabies in cattle is persistent in the country, with variations over time and that, if the surveillance measures do not change, the tendency is for the number of cases to remain constant in a short time. In addition, the variables that are considered to explain the occurrence of rabies cases in the literature and by the PNCRH have little or no influence, requiring future studies that aim to identify variables that better explain the occurrence of rabies cases in Brazil. The degree of spatial dependence and geographically weighted regression and the Gi * method of Getis-Ord of the variables were analyzed by the ArcGIS 10.3 software, the other statistical and transition matrix analyzes were performed using the software and R Studio 3.5.1, with the markovchain package. The spatial analysis showed a predominance of areas with fat content from 3.1 to 3.6g / 100g and areas with fat content from 3.6 to 4.2g / 100g. For the lactose content, a predominant area was observed with 4.32 to 4.45g / 100g and some areas with 4.46 to 4.54g / 100g. There was a low influence of altitude, rainfall and precipitation x altitude interaction on the fat, protein, lactose, total solids and defatted dry matter content in the studied area. Instituição: Universidade Federal Rural de Pernambuco Tipo do documento: Tese Mon, 22 Feb 2021 00:00:00 GMT http://www.tede2.ufrpe.br:8080/tede2/handle/tede2/8756 2021-02-22T00:00:00Z http://www.tede2.ufrpe.br:8080/tede2/handle/tede2/8755 Título: Análise espaço-temporal do número de casos de dengue em Pernambuco, Brasil: modelagem e métodos computacionais Autor: FREITAS, Jucarlos Rufino de Primeiro orientador: CUNHA FILHO, Moacyr Abstract: The present study aimed to analyze the Spatio-temporal distribution in the weekly dengue cases number for Pernambuco state, Brazil. Weekly records of the dengue cases number were collected, made available through the Citizen Information Service, and the historical series of weekly precipitation provided by the Pernambuco Agency for Water and Climate, in the period 2000 to 2018. To assess the correlation between the variables (dengue versus pecipitation), the Pearson correlation coefficient was calculated and the generalized linear models were proceeded. In the state of Pernambuco were confirmed 508,948 dengue cases from 2000 to 2018. The spatial distribution of cases showed a high concentration in the Metropolitan Mesoregion of Recife in the 19 years analyzed, with greater frequency from Recife municipality. There was formation of clusters with positive spatial autocorrelation between municipalities in the Metropolitan Mesoregion of Recife. Positive correlations were observed in 75.54% of the municipalities. The Negative Binomial model proves to be satisfactory to analyze the distribution of dengue cases. The computerized processing of georeferenced data allows identifying municipalities with greater vulnerabilities, providing subsidies for the regulatory agencies of epidemics and endemics in the state of Pernambuco to plan their actions. Instituição: Universidade Federal Rural de Pernambuco Tipo do documento: Tese Fri, 18 Feb 2022 00:00:00 GMT http://www.tede2.ufrpe.br:8080/tede2/handle/tede2/8755 2022-02-18T00:00:00Z http://www.tede2.ufrpe.br:8080/tede2/handle/tede2/8754 Título: Rapid detection approach for electronic nose systems using deep learning models Autor: GAMBOA, Juan Carlos Rodriguez Primeiro orientador: OLIVEIRA JUNIOR, Wilson Rosa de Abstract: The present thesis work focused on proposing a method to accelerate the identification of the specimens in electronic nose systems. Since the conventional data processing approach used in E-Nose is based on an initial stage of signal preprocessing applying techniques to perform the feature extraction (dynamic and static characteristics) for obtaining the odor fingerprint. Besides, in some cases, it is required to implement a feature selection method to choose the best attributes before the classification tasks using pattern recognition methods. For example, a Support Vector Machine (SVM) is one of the most common processing methods for odor recognition by the electronic olfactory systems. Therefore, the use of the traditional approach needs the whole measurement to obtain the main odorant parameters involving preprocessing techniques, which represents a challenge when aiming to perform real-time odors recognition. Thus, in this work is presented an approach for electronic olfactory systems data processing focused on the treatment of raw data based on a rising-window protocol to find an early portion of the sensor signals with the best recognition performance. We compared the proposed approach against a traditional method (using the entire response curves, applying preprocessing techniques to extract the features and later processing them using an SVM algorithm) in a real application with measures acquired with our developed system. Further, to validate the use of the proposed approach at different settings of electronic olfactory systems, we conducted more tests with several datasets and using deep learning techniques like convolutional neural network CNN. The results showed outperformance accuracy compared with the traditional approach with the advantage of using an early portion of the responses of the sensors, reducing the necessary time to make forecasts. Instituição: Universidade Federal Rural de Pernambuco Tipo do documento: Tese Tue, 18 Feb 2020 00:00:00 GMT http://www.tede2.ufrpe.br:8080/tede2/handle/tede2/8754 2020-02-18T00:00:00Z