Navegando por Assunto "Artificial neural networks"

Agora exibindo 1 - 20 de 25

Acesso aberto (Open Access)
Algoritmos para seleção de metodologias de avaliação de softwares educacionais
(Universidade Federal do Pará, 2023-09-26) CASTILHO, Janize Monteiro de; FARIAS, Fabricio de Souza; http://lattes.cnpq.br/1521079293982268; https://orcid.org/0000-0003-4344-6953
In order to assist the teaching-learning processes, many teachers have decided to use Educational Software (ES) in their classrooms. However, to choose a ES as a teaching resource it is essential to endorse the methodology used by the teacher, once it needs to be pedagogically and functionally appropriate to meet the needs and objectives present in the classroom. Also, it is necessary to use mechanisms that the ES endorses to verify its adequacy to the professor’s objectives. Currently, it is verified that there are various techniques and methodologies available in the literature for ES assessment, but there is still no solution for decision making and selection of a ES that fully addresses the profiles of users and their different needs to be met by certain methodological application, or that arises from demand originating from the development of solutions based on demand and with a low capacity for generalization in terms of practical application. In this way, solutions are available without standardization and that several times do not take into consideration criteria relating to quality, measurement scales and verification procedures of the ES. This heterogeneity makes the evaluation of an ES very difficult, since the subjectivity in the selection of ES evaluation methodology can produce inconclusive results. Given this context, this work created a quality model that considers 24 ES assessment methodologies available in the literature and aims to automate the selection of ES assessment methodology based on the application of artificial intelligence (AI) algorithms, reducing the possibility of subjectivity in the screening process. During the investigation we used Natural Language Processing (NLP), Random Forest, k-Nearest Neighbors and Artificiais Neurais Networks. In all research scenarios, the natural language algorithm was combined with other algorithms, offering a solution based on the application of hybrid and loosely coupled AI algorithms, with excellent results. In this way, simulations were carried out considering NLP+Random Forest, NLP+k-Nearest Neighbors and NLP+Artificial Neurais Networks. After the simulations, the results indicate that it is possible to determine the best ES assessment methodology using AI algorithms, with the best results obtained with the combination of NLP+Random Forest.
Acesso aberto (Open Access)
Avaliação de modelos de inteligência artificial híbridos na estimativa de precipitações
(Universidade Federal do Pará, 2022-03-18) GOMES, Evanice Pinheiro; BLANCO, Claudio José Cavalcante; http://lattes.cnpq.br/8319326553139808
The hydrological analyzes carried out from rainfall in the Amazon are essential due to its importance in climate regulation, regional and global atmospheric circulation. However, in this region, there are limitations related to data series with short periods and many flaws, especially in the daily scale. Despite significant advances in science and technology, practical and accurate predictions have been a major concern due to their complexity. Therefore, several conceptual models, empirical or hybrid, have been tested to forecast rain with greater precision. Among empirical models, those that incorporate artificial intelligence (AI) methods are potentially useful approaches to simulate the precipitation process. Artificial Neural Networks (ANN), as AI models, are able to establish a relationship between historical inputs (rain, flow, etc.) and the desired outputs, through a non-linear function composed of several factors that are adjusted to the observed data, allowing your prediction. Thus, to improve the precipitation analysis, hybrid models were developed, involving Artificial Neural Network (ANN) of the type with Time Delay (TDNN), ELMAN network, Radial Base network (RBF) and Adaptive Neuro-Fuzzy Inference System (ANFIS), coupled with Maximum Overlap Discrete Wavelet (MODWT). Six rainfall gauge station were adopted, which are located in different biomes of the region, and satellite data (CMORPH). Rainfall data were evaluated by seasonal periods (rainy and dry). The results obtained demonstrated that the MODWT-ANFIS model had the best capacity to simulate the daily precipitation of the evaluated rainfall gauge station, even for dry periods, which are known to be more difficult to be simulated in relation to the rainy periods. In this case, data entries lagged by 4 days and 5 days performed better, with Nash values close to 1.0 and root mean square errors below 0.001.
Acesso aberto (Open Access)
. Clusterização, classificação e predição de “pré-efeito anódico” de cuba eletrolítica de alumínio primário
(Universidade Federal do Pará, 2020-08-21) CONTE, Bruno Nicolau Magalhães de Souza; OLIVEIRA, Roberto Célio Limão de; http://lattes.cnpq.br/4497607460894318
The industrial sector is one of the main responsible for the serious environmental situation on the planet and also for increasing legal requirements, in relation to the waste generated. On the other hand, many companies have reacted proactively, based on the implementation of management strategies such as: clean production, environmental certification, reduction of toxic waste, recycling, sustainable consumption and reuse, mainly. It is worth mentioning that the aluminum reduction process is responsible for a large amount of greenhouse gas emissions and, thus, promotes environmental impacts and serious climate changes. During the aluminum reduction process, the occurrence of the anodic effect causes an extreme increase in the tub tension and, consequently, an increase in the bath temperature, with very high temperatures, resulting in a thermal disturbance, with the possibility of melting the insulating layer of the vat and the final consequences are the loss of production in the entire vat line, its shortened service life and the production of PFC gases. Seeking a strategy based on sustainability, I try to take into account the problem of the worsening of the Greenhouse Effect, the extreme increase in kiln tension and, consequently, the loss of production in the entire line of vats, this work proposes the use of an Artificial Neural Network together with Clustering algorithms to automatically create anodic pre-Effect labels, and thus predict the nonlinear dynamic behavior of the primary aluminum reduction industry oven anodic pre-effect, based on actual vat data electrolytic. With the use of these Machine Learning models, it is possible to predict the occurrence of the anodic pre-effect, allowing process operators to take mitigating measures to suppress the anodic effect, avoiding the loss of aluminum production in the vat and decreasing the emission of gases from the greenhouse effect.
Acesso aberto (Open Access)
Comparação entre regressão linear, redes neurais artificiais e árvores de regressão para quantificação do impacto harmônico de múltiplas cargas em redes elétricas de distribuição.
(Universidade Federal do Pará, 2018-11-19) PAIXÃO JÚNIOR, Ulisses Carvalho; TOSTES, Maria Emília de Lima; http://lattes.cnpq.br/4197618044519148
In recent years, the socio-economic development of the population, the growth of commercial and industrial sectors, as well as the ever-increasing installation of new electrical loads, have generated great evolution in demand of electricity consumption. In turn, to obtain more efficient systems, the manufacturers have produced equipment more energy efficient for residential, commercial and industrial use. However, these loads due their non-linearities, have contributed significantly to the increase in harmonic distortion levels of voltage and current, raising the concern of the power sector managers with respect to the power quality, mainly, due to the difficulty in the identification of the origin of the harmonic distortion. Therefore, to anticipate the harmonic effects and meet the current legislation, through computational techniques, this work emphasis is placed on the common coupling point (CCP) of consumers and utility, regardless of consumption characteristics and loads, to assess the harmonic impacts in his grid, besides comparing the reliability level of the techniques through the mean absolute error (MAE). The proposed methodology uses the Electrical Power Quality System (SISQEE) software that allows the use of three different computational techniques, such as Linear Regression, Artificial Neural Networks and Regression Trees, to evaluate the harmonic contribution of each feeder at the point of interest of the chosen electric grid. To prove the validity of the methodology, two case studies, based on real measurements at a university and at an industrial district, was carried out with a minimum sampling period of seven days using power quality analyzers, according to the distribution procedures by ANEEL (PRODIST). As a result of the power quality, it was verified how much each feeder impacts the voltage and current distortion at the CCP, besides classifying the feeders in relation to their respective impacts in the studied electrical grid. Also, as a result, the studies allowed the evaluation of performance between the different techniques, with different time intervals (weekly, daily and per load level), allowing to classify the behavior and reliability of each technique in each period. As a conclusion of the work, the proposed methods and analyzes presented allow managers to perform a more efficient mitigation action of the harmonic impacts caused in the electrical network and, also, to identify the differences between the techniques and their degree of reliability, in accordance with the time intervals studied.
Acesso aberto (Open Access)
Estimação da porcentagem de flúor em alumina fluoretada proveniente de uma planta de tratamento de gases por meio de um sensor virtual neural
(Universidade Federal do Pará, 2011-06-22) SOUZA, Alan Marcel Fernandes de; OLIVEIRA, Roberto Célio Limão de; http://lattes.cnpq.br/4497607460894318; AFFONSO, Carolina de Mattos; http://lattes.cnpq.br/2228901515752720
The industries have been often seeking to reduce operating expenses, as to increase profits and competitiveness. To achieve this goal, it must take into account, among other factors, the design and implementation of new tools that accurately, efficiently and inexpensively allow access to information relevant to process. Soft sensors have been increasingly applied in industry. Since it offers flexibility, it can be adapted to make estimations of any measurement, thus a reducing in operating costs without compromising the measurements, and in some cases even improve the quality of generated information. Since they are completely softwarebased, they are not subjected to physical damage as the real sensors, and are better adaptated to harsh environments with hard access. The success of this king of sensors is due to the use of computational intelligence techniques, which have been widely used in the modeling of several nonlinear complex processes. This work aims to estimate the quality of alumina fluoride from a Gas Treatment Center (GTC), which is the result of gaseous adsorption on alumina virgin, using a soft sensor. The model that emulates the behavior of a alumina quality sensor the plant was created using an artificial intelligence technique known as Artificial Neural Network. The motivations of this work are: perform virtual simulations without compromising the GTC and make accurate decisions based not only on the operator's experience, to diagnose potential problems before they can interfere with the quality of alumina fluoride; maintain the aluminum reduction pot control variables within normal limits, since the production from low quality alumina strongly affects the reaction of breaking the molecule that contains this metal. The benefits this project brings include: increasing the GTC efficiency, producing high quality fluoridated alumina and emitting fewer greenhouse gases into the atmosphere and increasing the pot lifespan.
Acesso aberto (Open Access)
Estimação das parcelas de contribuição de cargas não lineares na distorção harmônica de tensão de um barramento de interesse do sistema elétrico de potência utilizando rede neural artificial
(Universidade Federal do Pará, 2019-09-06) MANITO, Allan Rodrigo Arrifano; TOSTES, Maria Emília de Lima; http://lattes.cnpq.br/4197618044519148; BEZERRA, Ubiratan Holanda; http://lattes.cnpq.br/6542769654042813
This work presents a methodology to estimate the non-linear loads contribution on voltage harmonic distortion at a bus of interest in the electric power system. The estimation process is carried out through the development of a model based on artificial neural networks (ANN) added to a sensitivity analysis in neural network input. The ANN model input is constituted by the non-linear loads harmonic currents considered in the studied system, and the ANN output corresponds to the harmonic voltage values in the bus under study, for the same harmonic frequency. The study is carried out for each harmonic order individually and the data required for the construction of the model as well as for the results validation have been obtained from synchronized measurement campaigns and by computational simulation, using harmonic load flow studies. Comparisons between reference results through computational simulation with the results obtained by neural model are carried out and it is observed that the developed methodology is able to classify correctly the impact of non-linear loads in the voltage distortion at a bus of interest of the electric system. Additionally, the effectiveness of the methodology is tested in two real systems in order to verify the good performance of this methodology considering real data obtained during measurement campaigns.
Acesso aberto (Open Access)
Um framework para a previsão de cenários com o uso de sistemas híbridos neurogenéticos para compra e venda de energia elétrica no mercado futuro
(Universidade Federal do Pará, 2012-05-04) CONDE, Guilherme Augusto Barros; FRANCÊS, Carlos Renato Lisboa; http://lattes.cnpq.br/7458287841862567
In the context of time series forecasting, is great the interest in studies of forecasting methods of time series that can identify existing structures and patterns in historical data, allowing generate the next patterns of the series. The proposal defended in this thesis is the development of a framework that uses the full potential of forecasting techniques (neural networks) with the optimization techniques (genetic algorithms) in a hybrid system that well enjoy the advantages of each of these techniques to the generation of future scenarios that can show, in aaddition to normal forecasts based on historical values, alternative pathways of the curves of time series analyzed.
Acesso aberto (Open Access)
Handoff de espectro em redes baseadas em rádio cognitivo utilizando redes neurais artificiais
(Universidade Federal do Pará, 2011-02-03) BALIEIRO, Andson Marreiros; COSTA JÚNIOR, Carlos Tavares da; http://lattes.cnpq.br/6328549183075122
Nowadays, the radio spectrum is one of the most important natural resources in the world. According to the FCC (Federal Communications Commission) report, the licensed bands are abundant but poorly utilized. The Cognitive Radio (CR) technology aims improving the spectral efficiency through the opportunistic access to the electromagnetic spectrum. It enables that new applications based on wireless communications are supported, without causing interference at the licensed communication and trying to ensure the quality of service to the applications. Thus, the efficient spectrum handoff is a critical requirement to be taken into account for the success of CR systems. This work outlines concepts about the cognitive radio technology. It proposes and evaluates a proactive strategy based on Artificial Neural Networks for spectrum handoff in CR networks. The performance of the proposal regarding the inference level to the primary user, spectrum handoff number triggered by secondary user and spectral utilization is compared with that one obtained by a reactive scheme. Differently from previous works, this study considers measured traces made available by IEEE Dyspan 2008 in order to evaluate our proposal. Numerical results show the superiority of the proposed scheme.
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)
Implementação de modelos computacionais na predição temporal e espaço-temporal de parâmetros de qualidade de água
(Universidade Federal do Pará, 2021-12-14) ALMEIDA, Anderson Francisco de Sousa; MERLIN, Bruno; http://lattes.cnpq.br/7336467549495208; HTTPS://ORCID.ORG/0000-0001-7327-9960; GONZÁLEZ, Marcos Tulio Amaris; http://lattes.cnpq.br/9970287865377659
The quality of water is directy related to is level of pollution caused by anthropic and industrial actions, with a consequent reduction in the availability of quality water. Therefore, limological monitorig of the basic parameters os water quality is carried out, as away of obtaining data that guide the decision-making of water resouces management bodies. In this context, the present study has the implementation of machine learning algorithms to predict temporal and spatiotemporal water quality parameter data. The ML techniques used were linear regression, ramdom forest, MLP and LSTM neural networks. Two collection points from a Water Resources Management Unit in São Paulo, Brazil were used. Models are evaluated using MAPE( mean absolute percentage eror) and RMSE( root mean squared erro) metrics. Therefore, in temporal prediction, the LSTM technique presented the best performace in relation to the other techniques and the data used, as it has the lowest average RMSE result, with 2.47. However, in spatiotemporal prediction, MLP has the best performace both in relation to the other techniques and the data used , as it has the lowest averagee results of MAPE and RMSE, respectively, 5.94% and 1.34. Thus, these performaces of neural networks can be justified by the non-linearity of the parameter data. Other than that, the results of the experiments aim to contribute to the water quality monitorng process and assist in the planning of water management, so that it meets current legislation and enales the indication of public policies, through machine learning models in prediction of water quality parametes.
Acesso aberto (Open Access)
Localização de falhas em linhas de transmissão utilizando decomposição harmônica e redes neurais artificiais
(Universidade Federal do Pará, 2006-07-05) GOMES, Cristiane Ruiz; VIEIRA JÚNIOR, Petrônio; http://lattes.cnpq.br/1958791286192330
This work proposes a new methodology for fails location in transmission lines. This methodology consists in using harmonic decomposition of the leakage current and in the application of an Artificial Neural Network (ANN), this is able to recognize patterns of normal and fails conditions of a transmission line. It was developed a Pi model capable to use real data of voltage and current of the three phases. In this model values of capacitance, inductance and resistance can be modified in agreement of weather conditions. Fails were generated in all the towers with different values of capacitance. The input/output data were used to train the neural network. The real voltage and current data acquisition were done by instruments installed in the two terminals of the Guamá-Utinga transmission line belonging to Centrais Elétricas do Norte - ELETRONORTE. The computation of the parameters was made by the well known matricial method and was improved by Finite Element Method. An ANN was developed with Matlab software. For training the ANN it was used the backpropagation resilient algorithm, which presented good performance, been fast and accurate. The ANN was trained by two different sets to analyze differences between outputs. In the two cases results were satisfactory.
Acesso aberto (Open Access)
Metodologia baseada em modelo teórico para avaliação do estado operativo de linha de transmissão
(Universidade Federal do Pará, 2015-05-15) NEGRÃO, Martin Max Luis de Castro; BARREIROS, José Augusto Lima; http://lattes.cnpq.br/1246564618922453; VIEIRA JÚNIOR, Petrônio; http://lattes.cnpq.br/1958791286192330
Conventionally monitoring operating conditions of a power transmission line is accomplished by periodic inspections along this line. This monitoring allows corrective maintenance by finding faults during the inspection. But in more efficient maintenance, predictive techniques that are characterized by real-time monitoring should be employed. The predictive techniques verify the operating status using normal function models for fault detection and fault models for the diagnosis to be employed in PDI (Fault Detection and Isolation). So, we developed a mathematical model appropriate for application to predictive maintenance of transmission line segments at low cost, without the need for sensors distributed along the line. This model allows the use of the methodology for detecting faults (PDI) by monitoring the leaka ge current of transmission lines. The use of the model also allows obtaining a new indicator of the condition of normal and abnormal functioning of a transmission line, which is the capacitance of harmonic frequencies. The model was validated through measurements obtained on a section of transmission line, by means of an artificial neural network.
Acesso aberto (Open Access)
Metodologia de predição de perda de propagação e qualidade de vídeo em redes sem fio indoor por meio de redes neurais artificiais
(Universidade Federal do Pará, 2018-02-27) CRUZ, Hugo Alexandre Oliveira da; CAVALCANTE, Gervásio Protásio dos Santos; http://lattes.cnpq.br/2265948982068382
This dissertation presents a methodology that aims to assist the planning of indoor wireless network systems, which require prior knowledge of the environments in which they will be deployed. Thus, accurate signal analysis is necessary by means of a statistical empirical approach, which takes into account some factors that influence the propagation of the indoor signal: architecture of the buildings; arrangement of furniture inside the compartments; numbers of walls and floors of various materials, and the spread of radio waves. The methodology adopted is based on measurements with a cross-layer approach, which demonstrates the impact of the physical layer in relation to the application layer, in order to predict the behavior of the Quality of Experience (QoE) metric, called Peak signal- to-noise ratio (PSNR), in 4K video streams on 802.11ac wireless networks in the indoor environment. In order to do so, measurements were performed, which demonstrate how the signal / video degrades in the studied environment. It is possible to model this degradation by means of a computational intelligence technique, called Artificial Neural Networks (RNA), in which input parameters are inserted as, for example, the distance from the transmitter to the receiver and the number of walls crossed in order to predict loss of propagation and loss of PSNR. In order to evaluate the predictive capacity of the proposed methods, the values of the Root Mean Sqare (RMS) errors between the measured and predicted data were obtained by the prediction methods loss of propagation and loss of PSNR, with respective values of 2.17 dB and 2.81 dB.
Acesso aberto (Open Access)
Metodologia integrada utilizando sensoriamento remoto em redes neurais artificiais na quantificação do potencial de biomassa florestal na Amazônia
(Universidade Federal do Pará, 2008-04-08) ALMEIDA, Arthur da Costa; ROCHA, Brigida Ramati Pereira da; http://lattes.cnpq.br/9943372249006341
Pattern recognition and pattern classification in digital images is a very important skill, today. With them, it is possible to recognize and identify target objects in those images. This work proposes an integrated methodology for pattern recognition related to biomass in the Amazon tropical rainforest to extract information about bioenergetics potential for electric energy production for use with isolated Amazonian communities. To achieve this aim, information gathered about forest inventory was mixed with pattern classification and recognition in medium resolution satellite imagery such as those from LANDSAT and CBERS. The approach used in this work comes from the computational intelligence area, using artificial neural networks equipped with radial basis functions and Kohonen´s self organizing maps. The results serve as input to a geographical information system application which creates and manages a geographical database for energetic planning with renewable energy resources applicable to isolated Amazonian communities.
Acesso aberto (Open Access)
Metodologia para compressão de sinais de energia elétrica a partir de registros de forma de onda utilizando algorítmos genéticos e redes neurais artificiais
(Universidade Federal do Pará, 2016-12-16) BARROS, Fabíola Graziela Noronha; NUNES, Marcus Vinícius Alves; http://lattes.cnpq.br/9533143193581447; BEZERRA, Ubiratan Holanda; http://lattes.cnpq.br/6542769654042813
This thesis proposes a methodology for compression of electrical power signals from waveform records in electric systems, using genetic algorithm (GA) and artificial neural network (ANN).The genetic algorithm is used to select and preserve the points that better characterize the waveform contoursA and the artificial neural network is used in the compression of other points as well as on the signal reconstruction process. Thus, the data resulting are formed by a part of the original signal and by a compressed complementary part in the form of synaptic weights. The proposed methodology selects and preserves a percentage of the original signal samples, which are aspects not explored in the literature. The method was tested using field data obtained from an oscillographic recorder installed in a 230kV electrical power system. The results presented compression rates ranging from 88.36 to 95.86 for preservation rates ranging from 2.5 to 10 , respectively.
Acesso aberto (Open Access)
Modelagem neural da resistência elétrica dos fornos de redução do alumínio
(Universidade Federal do Pará, 2015-10-16) CONTE, Thiago Nicolau Magalhães de Souza; OLIVEIRA, Roberto Célio Limão de; http://lattes.cnpq.br/4497607460894318
The paper evaluates two types of Artificial Neural Networks to model dynamically the behaviour of the electrical resistance of a primary aluminum reduction furnace. The proposal is to use Direct Multilayer neural networks (RNMD) and Recurrent Neural networks (RNR) to model the electrical resistance of the oven. For each of these Neural Networks is explored its ability to model dynamic systems, either by varying the number of layers of neurons, as well as the number of neurons in each layer, varying the neural network input signals, etc. The data to be used in modeling from a Brazilian factory of primary aluminum. This modeling can be used to control the distance (up or down) between the electrodes, anodes and cathodes of the reduction that it consists primarily of carbonaceous materials. In this way the system of control has the task of maintaining the value of resistance within acceptable ranges of operation always attempting to ensure thermal stability and consequently the production of primary aluminum, high-purity, based on data available online in the control system of the plant. Through these electrodes are injected electrical currents keep that, besides the electrolysis itself cause the electrolytic bath, raising its temperature to a range up to 960° C. The motivation for the work is in high complexity of primary aluminum reduction process, whose nature is non-linear and the same suffering directly related variables influence the dynamics of the process, often imperceptible process engineers from the factory, but can be perceived by means of computational intelligence techniques reflecting about the different operating conditions of the real system.
Acesso aberto (Open Access)
Modelo de previsão hidrológica utilizando redes neurais artificiais: um estudo de caso na bacia do Rio Xingu- Altamira-Pa
(Universidade Federal do Pará, 2019-10-10) SILVA, Arilson Galdino da; CASTRO, Adriana Rosa Garcez; http://lattes.cnpq.br/5273686389382860
Knowledge about the extent of riverbed overflow is extremely necessary for the determination of areas at risk. The City of Altamira-PA, located on the banks of the Xingu River, historically suffers from extreme events of floods that provoke floods, causing great damages to the population. Considering the problem, this paper presents a monthly level prediction system of the Xingu River based on neural networks perceptron of multiple layers. For the development of the system, rainfall data were used in the basin and sub-basins of the Xingu River, and SST information (Sea Surface Temperature) from 1979 to 2016. The Satisfactory results demonstrate the great applicability of Artificial Neural Networks to the flood prediction problem, as compared to other methodologies have greater precision in finding solutions for nonlinear problems. For the treatment and selection of the input variables, the correlation approach was used, with the objective of improving the accuracy of the results, thus selecting the best information with their respective lags, in which they are inserted in three prediction scenarios: model with rainfall data, model with sea surface temperature information and application using the SST junction with rainfall. To measure the prediction capacity of the proposed methods, the Mean Squared Error (MSE) and coefficient of determination (R²) values were obtained for the best strategy, using only oceanic variables, SST, being the values 2,99x104 and 0,9991 considering, mainly, the treatment of input values of the Neural Network.
Acesso aberto (Open Access)
Modelo híbrido baseado em séries temporais e redes neurais para previsão da geração de energia eólica
(Universidade Federal do Pará, 2018-08-30) ALENCAR, David Barbosa de; OLIVEIRA, Roberto Célio Limão de; http://lattes.cnpq.br/4497607460894318; AFFONSO, Carolina de Mattos; http://lattes.cnpq.br/2228901515752720
The electric power generation through wind turbines is one of the practically inexhaustible alternatives sources of electric power. It is considered a source of clean energy, but still requires a lot of research to develop science and technologies that ensure uniformity in generation, providing a greater participation of this source in the energy matrix in Brazil as in the world, because the wind presents abrupt variations speed, density, and other important variables. In wind-based electrical systems, each forecast horizon is applied to a specific segment, forecast of minutes, hours, weeks, months, and future years of wind behavior, in order to evaluate the availability of energy for the next period, relevant information in the dispatch of the generating units and in the control of the electric system. This thesis aimed to develop ultra-short, short, medium and long-term prediction models of wind speed, based on computational intelligence techniques, using Artificial Neural Networks, SARIMA models and hybrid models and to predict the generation capacity of power for each horizon. For the application of the methodology, the meteorological variables of the database of the national environmental data system SONDA, Petrolina station, were used for the period from January 1st, 2004 to March 31st, 2017. The performance of the models was compared with 5, 10 and 20 steps forward, considering minutes, hours, days, weeks, months and years as the forecast horizon. The hybrid model obtained better response in the forecasts, among which the hour horizon was highlighted.
Acesso aberto (Open Access)
Modelos para previsão de carga a curto prazo através de redes neurais artificiais com treinamento baseado na teoria da informação
(Universidade Federal do Pará, 2011-11-04) ALVES, Wesin Ribeiro; CASTRO, Adriana Rosa Garcez; http://lattes.cnpq.br/5273686389382860
The previous knowledge of the load value is almighty important to the electric power system planning and operation. This paper presents results of an investigative study of application of Artificial Neural Networks as a Multilayer Perceptron with the training based on Information Theory to the problem of short term load forecasting. The learning based on Information Theory focuses on the use of the amount of information (Entropy) for the training of neural network. Two forecaster models are presented, and that they was developed using real data from an energy utility. To compare and verify the efficiency of the proposed systems, it was also developed a forecasting system using neural network trained based on the traditional criterion of mean square error (MSE). The results has showed the efficiency of proposed systems, which had better results when compared with the forecasting system based on neural network trained by criterion of MSE and with forecasting system already was presented in the literature.
Acesso aberto (Open Access)
Nova metodologia para análise e síntese de sistemas de aterramento complexos utilizando o método lN-FDTD, computação paralela automática e redes neurais artificiais
(Universidade Federal do Pará, 2008-02-29) OLIVEIRA, Rodrigo Melo e Silva de; SOUZA SOBRINHO, Carlos Leônidas da Silva; http://lattes.cnpq.br/1450994881555781
The FDTD method in General Coordinates (LN-FDTD) was implemented for analyzing structures not coincident to the Cartesian coordinate system. The method solves the Maxwells equations in time domain, allowing the calculation of data concerning the transitory and steady-state responses of such structures. The method is applied to analyze of special grounding electrodes. A new formulation for the truncating technique UPML, for conductive media, referred here as LN-UPML, was developed and implemented in order to make the simulations viable. A new methodology for locating grounding grid faults using two Artificial Neural Networks is presented. The LN-FDTD software was tested and validated though simulations of various grounding systems. A graphical user interface, named LANE SAGS, was implemented to simplify the use and to automate the data processing.