Navegando por Assunto "Redes neurais (Computação)"

Agora exibindo 1 - 16 de 16

Acesso aberto (Open Access)
Algoritmo genético retroviral iterativo
(Universidade Federal do Pará, 2010-09-10) MOREIRA, Renato Simões; AFFONSO, Carolina de Mattos; http://lattes.cnpq.br/2228901515752720
This work presents the development of a hybrid meta-heuristic based on the viral life cycle, specifically from Retroviruses, which are part of nature’s swiftest forms. This algorithm is called Retroviral Iterative Genetic Algorithm (RIGA) and uses as computational basement Genetic Algorithm (GA) and biological basement retroviral replication characteristics, which provides a great diversity increasing the probability to find the solution, what is confirmed by better results obtained by AGRI than AG.
Acesso aberto (Open Access)
Aplicação de redes neurais artificiais na classificação de padrões filogeográficos com base na variabilidade genômica do DNA mitocondrial
(Universidade Federal do Pará, 2007-12-20) GOMES, Larissa Luz; SANTOS, Ândrea Kely Campos Ribeiro dos; http://lattes.cnpq.br/3899534338451625; OLIVEIRA, Roberto Célio Limão de; http://lattes.cnpq.br/4497607460894318
Historically, the process of formation of the populations of the Amazon, as well as from all Brazilian territory, involved three main ethnic groups: the Amerindian, European and African. As a result, these populations have in general admixed constitution the point of view of social and biological. Since the end of the last century, studies of mitochondrial DNA (mtDNA) has been developed for the purpose of estimating the mixture inter present in these populations. For this, it is of fundamental importance classification of a particular strain of mtDNA in one of more than 250 haplogroups/sub-clades proposed in the literature. With the goal of developing an automated system, precise and accurate classification of the sequences (strains) of mtDNA, this has worked hand of the art of Artificial Neural Networks (RNAs) on the basis of the studies of Philogeography. For this classification, four networks have been developed artificial neural direct, with multiple layers and the learning algorithm to backpropagation. The entries of each network equivalent positions at nucleotide polymorphic region's hipervariable of mitochondrial DNA, which returned as output classification specific to each lineage. Subsequent to the training, all the networks had indices of adjustment of 100%, demonstrating that the technique of Artificial Neural Network (ANN) can be used, with success, in the classification of standards Philogeography based on mitochondrial DNA.
Acesso aberto (Open Access)
Cálculo de porosidade com a rede neural competitiva
(Universidade Federal do Pará, 2015-10-26) ROSELLÓN GUZMAN, Laura Yesenia; ANDRADE, André José Neves; http://lattes.cnpq.br/8388930487104926
Porosity is the petrophysical property that quantifies the fluid volume in the reservoir rock under for subsurface original condition. However, its calculation by the densityneutron method is extremely difficult in non cored borehole by the lack of the knowledge about the matrix physical properties (density and neutron porosity). This work presents a method for enabling the use of density-neutron Method in non cored boreholes, showing a realistic estimate of the matrix physical properties for each reservoir layer, using a angular competitive neural network. For each layer, network training is performed in the density-neutron plot built with the points of this layer and the information about the grain density (matrix density), obtained in the core analysis. This method is presented with synthetic data, which satisfy the petrophysical model and real data from two cored wells in the Namorado field, Campos basin.
Acesso aberto (Open Access)
Classificação de regiões de desmatamento via imagens do satélite landsat no nordeste do Pará
(Universidade Federal do Pará, 2023-12-18) CANAVIEIRA, Luena Ossana; COSTA, João Crisóstomo Weyl Albuquerque; http://lattes.cnpq.br/9622051867672434
Acesso aberto (Open Access)
Comparação dos algoritmos C4.5 e MLP usados na avaliação da segurança dinâmica e no auxílio ao controle preventivo no contexto da estabilidade transitória de sistemas de potência
(Universidade Federal do Pará, 2013-03-06) OLIVEIRA, Werbeston Douglas de; BEZERRA, Ubiratan Holanda; http://lattes.cnpq.br/6542769654042813; VIEIRA, João Paulo Abreu; http://lattes.cnpq.br/8188999223769913
This work compares the C4.5 and multilayer perceptron (MLP) algorithms applied for dynamic security assessment (DSA) and power system stability transient preventive control design. C4.5 is an algorithm of the decision tree (DT) technique and the MLP is a member of artificial neural network (ANNs) family. The advent of DTs and ANNs provides solution to real-time DSA issues in order to identify quickly when a power system is subjected to a critical disturbance (short-circuit) that may lead to transient instability. In addition, the knowledge obtained by both techniques can be utilized in the preventive control design to restore the power system security against critical disturbances. Based on the data base generation with exhaustive time-domain simulations, some specific critical disturbances are taken as examples to compare the C4.5 and MLP algorithms employed to DSA and guideline to preventive actions. The comparative study is tested on the New England power system. In the case studies, the knowledge database is generated by using PSTv3 (Power System Toolbox) software. The DTs and ANNs are trained and tested by the Rapidminer software. The obtained results have demonstrated a promising application of the C4.5 and MLP algorithms used in power system DSA and preventive control design.
Acesso aberto (Open Access)
Deconvolução de perfis de poço através de rede neural recorrente
(Universidade Federal do Pará, 2006-03-05) RUÉLA, Aldenize de Lima; ANDRADE, André José Neves; http://lattes.cnpq.br/8388930487104926
For oil industry, the logs analysis is the main information source about the presence and quantification of hydrocarbon in subsurface. However, in two situations the new logging technologies are not economically viable and conventional logging tools must be used: The reevaluation of mature oil fields and evaluation of marginal oil fields. In conventional logs its data acquisition procedure may blur the value of physical property and the vertical limits of a rock layer. We are talking about an old problem in well logging – The paradox between vertical resolution and depth of investigation of a logging tool. Nowadays it is well handling by the high technology of new tools, but this problem persists in conventional old tools, e.g. natural gamma ray log (GR). Here, we present a method to smooth this kind of linear distortion in well logs by an integration of classical well log convolution model with recurrent neural networks. We assume that a well log can be well represented by an in depth convolution operation between the variation of rock physical property (ideal log) and a function that causes the distortion, called as vertical tool response. Thus, we develop an iterative data processing, which acts as a deconvolution operation, composed by three recurrent neural networks. The first one seeks to estimate the vertical tool response; the second one search for the vertical limits definition of each rock layer and the last one is constructed to estimate the actual physical property. To start this process we supply an appropriated first guess of ideal log and vertical tool response. Finally, we show the improvements in vertical resolution and in the physical property evaluation produced by this methodology in synthetic logs and actual well log data from Lagunillas formation, Maracaibo basin, Venezuela.
Acesso aberto (Open Access)
Detecção de refletores sísmicos por rede neural discreta
(Universidade Federal do Pará, 1999) FERREIRA, Alexandre Beltrão; ANDRADE, André José Neves; http://lattes.cnpq.br/8388930487104926; LEITE, Lourenildo Williame Barbosa; http://lattes.cnpq.br/8588738536047617
The artificial neural networks have proven to be a powerful technique in solving a wide variety of optimization problems. In this work, we develop a new recurrent network, with no self-feedback loops, and no hidden neurons, for seismic signal processing, where this neural network gives the true polarity, reflectors location and magnitude estimations. The main characteristic of this neural network is the use of a type of activation function which permits three possible states of neurons, to estimate the position of the seismic reflectors in such way to reproduce its true polarities. The basic idea of this new neural network type, denominated here by discrete neural network (DNN), is to relate a cost function, that describes the geophysical problem, with the Liapunov function, that describes the neural network dynamics. In this way, the dynamics of the network leads to a local minimization of the Liapunov function, and will consequently lead to a minimization of the cost function. Thus, with a convenient output signal codification of the neural network a geophysical problem solution is obtained. The operational evaluation of this neural network architecture is performed with synthetic data obtained through the simple convolutional model and seismic ray theory, and its behavior explained with additive noise in the data with minimum, maximum and mixed phase time source pulses.
Acesso aberto (Open Access)
Estabilizador neural não-linear para sistemas de potência treinado por rede de controladores lineares
(2006-06) BARREIROS, José Augusto Lima; FERREIRA, André Maurício Damasceno; BARRA JUNIOR, Walter; COSTA JÚNIOR, Carlos Tavares da; BAYMA, Rafael Suzuki
Power System Stabilizers (PSS) have been applied as the most common solution to damp small magnitude and low frequency oscillations in modern electric power systems. Conventional Stabilizers, with fixed structure and parameters, have been used with this objective for several decades, but there are some system operation conditions where the performance of these linear stabilizers may deteriorate, especially when compared with that of stabilizers designed using modern control techniques. A Neural PSS, trained with a set of local linear controllers, is applied to establish the regions where a Conventional PSS shows low performance. Using non-linear digital simulations of a synchronous machine connected to an infinite-bus system and a multi-machine power system the Neural PSS is assessed showing superiority in those regions.
Acesso aberto (Open Access)
Identificação de fácies em perfis com rede neural direta
(Universidade Federal do Pará, 2015) GOMES, Kivia do Carmo Palheta; ANDRADE, André José Neves; http://lattes.cnpq.br/8388930487104926
The application of coring techniques is usually carried out in a limited number of vertical wells drilled in an oil field, causing the rarefaction of facies descriptions and not allowing a realistic characterization of reservoirs. Increased production of hydrocarbons in an oil field is extremely important for the oil industry and deeply dependent on the knowledge of the reserves in accordance with their petrophysical properties, which vary depending on geological facies. A better description of facies may reflect more realistic estimates of hydrocarbon volumes. This dissertation presents an intelligent algorithm capable of producing the transport of geologic information produced by the facies analysis of cores to the non-cored wells in an oil field, through the design of a direct neural network trained to perform a mapping of geological information in terms of the physical properties registered in the well logs. The intelligent algorithm processes the result produced by the neural network through a depth coherence filter to locate the boundaries of the layers along the well trajectory. For all of our cases the intelligent algorithm presented results compatible with the core analysis and independent of the size of the training set.
Acesso aberto (Open Access)
Modelagem chuva-vazão em bacias hidrograficas com suporte em redes neurais artificiais
(Universidade Estadual de Campinas, 1999-12-06) BARP, Ana Rosa Baganha; BARBOSA, Paulo Sérgio Franco; http://lattes.cnpq.br/9653654803297649
This work describes the use of two hydrological simulation deterministic models to represent the rainfuJI-runnoff processo The Itapetiniga, Almas and Guarapiranga rivers basin, located in the State of São Paulo and the Guaporé river basin located in the State of Mato Grasso, are taken as a case study. Both deterministic models used parameters optimization, with a nonlinear and unconstrained structure: (a) SMAP - Soil Moisture Accounting Procedure, which uses a first order optimization procedure; (b) Artificial Neural Network (ARN) model, which uses a second order optimzation procedure. Both models assume a montly interval to account rainfall and river flow. Some tests include a mix structure between SMAP and ARN, aiming at an evaluation of ARN potential to replace physical parameters and typical processes of conceptual rainfall-runnoff models.
Acesso aberto (Open Access)
Predição de comportamento de usuários oriundos do marketing digital por meio de redes neurais artificiais e aprendizado supervisionado
(Universidade Federal do Pará, 2019) ALVES, Vitor Pinheiro; TEIXEIRA, Otávio Noura; http://lattes.cnpq.br/5784356232477760; https://orcid.org/0000-0002-7860-5996
Success in attracting customers using marketing techniques creates a billionare problem wich is one of the most difficult that is selecting among the many prospects, which are more likely to become a customer. This work uses artificial neural networks to analyze the dataset generated from digital marketig techniques and classify which prospects have a greater chance to become a customer and which ones should be discarded. The Neural Network scores approximately 70% of cases among 3,541 records processed.
Acesso aberto (Open Access)
Reconhecimento de fáceis em perfis geofísicos de poços com rede neural competitiva
(Universidade Federal do Pará, 2015-02-27) COSTA, Jéssica Lia Santos da; ANDRADE, André José Neves; http://lattes.cnpq.br/8388930487104926
The description of a depositional system based on the recognition of sedimentary facies is critical to the oil industry to characterize the petroleum system. In the absence of these facies description by cores or outcrop, we present a methodology based on intelligent algorithm able to identify facies of interest in wireline logs. This methodology uses a competitive neural network to extract geological information from the physical properties mapped in the M-N plot. The competition among neurons identifies the facies of interest, which have been previously identified in a cored borehole in other non-cored boreholes in the same oil field. The purpose of this methodology is to encode and transmit the geological information gained in cored boreholes to non-cored wells and thus achieve the geological interpretation of the facies of interest in an oil field. This methodology has been evaluated with synthetic data and actual wireline logs from two cored boreholes drilled in the Namorado oil field, Campos Basin, Brazil.
Acesso aberto (Open Access)
Redes neurais convolucionais aplicadas à inspeção de componentes do vagão ferroviário
(Universidade Federal do Pará, 2020-02-03) ROCHA, Rafael de Lima; GOMES, Ana Claudia da Silva; http://lattes.cnpq.br/9898138854277399; SILVA, Cleison Daniel; http://lattes.cnpq.br/1445401605385329; https://orcid.org/0000-0001-8280-2928
The railcar is one of the most important assets in a mining company, where tons of ore are transported daily by it, besides, the railcar can be used to transport people. Therefore, the inspection of defects in structural components of the railcar is a very important activity, making it possible to avoid problems in railway logistics, as well as to prevent accidents. The inspection task is performed visually by an operating technician who is exposed to accidents where the inspection is performed, in addition to the possibility of human error due to stress, fatigue, and others. The pad is a rail component analyzed in this work, where it is responsible for the primary suspension, a role that is important in the railcar dynamics. Thus, the purpose of this work is to use deep learning techniques, specifically convolutional neural networks (CNN) for the component inspection. CNN classifies the image of the structural component analyzed concerning the possible state it is in the railway, absent pad, undamaged pad, and damaged pad. Also, it intends to investigate the contribution of the component image in the frequency domain obtained through the magnitude and phase of the discrete Fourier transform (DFT) of the original image (spatial domain) in the CNN classification process. Histogram equalization and increasing the number of images through data augmentation techniques are also examined to evaluate their collaborations in improving classification performance. The results of CNN inspection of the pad prove to be quite inspiring, especially when the spatial component image is used together with the DFT magnitude image of the original image as CNN inputs, which are superior when only the original (spatial) image of the component is used, achieving a classification accuracy of 95.65%. In particular, the method that uses the increase in the number of training images by the data augmentation and the spatial domain and frequency (magnitude) images achieves the highest accuracy, with 97.47%, which represents approximately 385.5 correctly classified images from a total of 395.2 images.
Acesso aberto (Open Access)
Redes neurais diretas e recorrentes na previsão do preço de energia elétrica de curto prazo no mercado brasileiro
(Universidade Federal do Pará, 2016-11-11) PEREIRA JUNIOR, Flaviano Ramos; OLIVEIRA, Roberto Célio Limão de; http://lattes.cnpq.br/4497607460894318
There are few articles about short term electricity price prediction in the Brazilian market. Existing works use ARIMA predictors and feedforward neural networks however, without input selection or lag selection for these inputs. Besides, there is no work with use of recurrent neural networks in the Brazilian electricity market. The short term electricity market may show important opportunities for the agents acting as the commercialization in this market is less bureaucratic in relation to the long-term market.. This article shows the use of feedforward and recurrent neural networks (besides comparison with the ARIMA model) to predict short term electricity price with the use of correlation for exogenous input selection for the networks and also for lag selection to these inputs. It is shown that, for one step forward predictions, both implemented networks outperforms the ARIMA model, and in general, feedforward network works better than recurrent network. Besides, lag selection in the input improves feedforward network performance.
Acesso aberto (Open Access)
Sistema hidrológico para previsão de risco na Amazônia utilizando redes neurais.
(Universidade Federal do Pará, 2019-03-02) PERES, Victor da Cruz; ROCHA, Edson José Paulino da; http://lattes.cnpq.br/2313369423727020
The estimation of the future behavior of the levels of a river basin is fundamental for the elaboration of the plan of management of its water resources. The objective of this research was to model the relationship between rainfall and level through a technique known as artificial neural networks (RNA). RNAs are empirical models with functions similar to the functioning of the human brain. In this research, the ability of RNA to model the rain-level process on a daily basis was evaluated. Influences of network architecture, initialization of weights, and extension of data series were considered during RNA training. The five RNAs that produced the best results were confronted with the observed results. The results were very satisfactory. Finding in a dry and full alert system in Itaituba-Pa.
Acesso aberto (Open Access)
Stormsom: clusterização em tempo-real de fluxos de dados distribuídos no contexto de BigData
(Universidade Federal do Pará, 2015-08-28) LIMA, João Gabriel Rodrigues de Oliveira; CARDOSO, Diego Lisboa; http://lattes.cnpq.br/0507944343674734; SANTANA, Ádamo Lima de; http://lattes.cnpq.br/4073088744952858