Desenvolvimento de uma arquitetura em hardware prototipada em FPGA para aplicações genéricas utilizando redes neurais artificiais embarcadas

This work proposes hardware architecture, VHDL described, developed to embedded Artificial Neural Network (ANN), Multilayer Perceptron (MLP). The present work idealizes that, in this architecture, ANN applications could easily embed several different topologies of MLP network industrial field. The MLP topology in which the architecture can be configured is defined by a simple and specifically data input (instructions) that determines the layers and Perceptron quantity of the network. In order to set several MLP topologies, many components (datapath) and a controller were developed to execute these instructions. Thus, an user defines a group of previously known instructions which determine ANN characteristics. The system will guarantee the MLP execution through the neural processors (Perceptrons), the components of datapath and the controller that were developed. In other way, the biases and the weights must be static, the ANN that will be embedded must had been trained previously, in off-line way. The knowledge of system internal characteristics and the VHDL language by the user are not needed. The reconfigurable FPGA device was used to implement, simulate and test all the system, allowing application in several real daily problems

Uma arquitetura para sistemas supervisórios industriais e sua aplicação em processos de elevação artificial de petróleo

The using of supervision systems has become more and more essential in accessing, managing and obtaining data of industrial processes, because of constant and frequent developments in industrial automation. These supervisory systems (SCADA) have been widely used in many industrial environments to store process data and to control the processes in accordance with some adopted strategy. The SCADA s control hardware is the set of equipments that execute this work. The SCADA s supervision software accesses process data through the control hardware and shows them to the users. Currently, many industrial systems adopt supervision softwares developed by the same manufacturer of the control hardware. Usually, these softwares cannot be used with other equipments made by distinct manufacturers. This work proposes an approach for developing supervisory systems able to access process information through different control hardwares. An architecture for supervisory systems is first defined, in order to guarantee efficiency in communication and data exchange. Then, the architecture is applied in a supervisory system to monitor oil wells that use distinct control hardwares. The implementation was modeled and verified by using the formal method of the Petri networks. Finally, experimental results are presented to demonstrate the applicability of the proposed solution

Análise de desempenho da rede neural artificial do tipo multilayer perceptron na era multicore

Artificial neural networks are usually applied to solve complex problems. In problems with more complexity, by increasing the number of layers and neurons, it is possible to achieve greater functional efficiency. Nevertheless, this leads to a greater computational effort. The response time is an important factor in the decision to use neural networks in some systems. Many argue that the computational cost is higher in the training period. However, this phase is held only once. Once the network trained, it is necessary to use the existing computational resources efficiently. In the multicore era, the problem boils down to efficient use of all available processing cores. However, it is necessary to consider the overhead of parallel computing. In this sense, this paper proposes a modular structure that proved to be more suitable for parallel implementations. It is proposed to parallelize the feedforward process of an RNA-type MLP, implemented with OpenMP on a shared memory computer architecture. The research consistes on testing and analizing execution times. Speedup, efficiency and parallel scalability are analyzed. In the proposed approach, by reducing the number of connections between remote neurons, the response time of the network decreases and, consequently, so does the total execution time. The time required for communication and synchronization is directly linked to the number of remote neurons in the network, and so it is necessary to investigate which one is the best distribution of remote connections

Simulador computacional para poços de petróleo com método de elevação artificial por bombeio mecânico

This work proposes a computer simulator for sucker rod pumped vertical wells. The simulator is able to represent the dynamic behavior of the systems and the computation of several important parameters, allowing the easy visualization of several pertinent phenomena. The use of the simulator allows the execution of several tests at lower costs and shorter times, than real wells experiments. The simulation uses a model based on the dynamic behavior of the rod string. This dynamic model is represented by a second order partial differencial equation. Through this model, several common field situations can be verified. Moreover, the simulation includes 3D animations, facilitating the physical understanding of the process, due to a better visual interpretation of the phenomena. Another important characteristic is the emulation of the main sensors used in sucker rod pumping automation. The emulation of the sensors is implemented through a microcontrolled interface between the simulator and the industrial controllers. By means of this interface, the controllers interpret the simulator as a real well. A "fault module" was included in the simulator. This module incorporates the six more important faults found in sucker rod pumping. Therefore, the analysis and verification of these problems through the simulator, allows the user to identify such situations that otherwise could be observed only in the field. The simulation of these faults receives a different treatment due to the different boundary conditions imposed to the numeric solution of the problem. Possible applications of the simulator are: the design and analysis of wells, training of technicians and engineers, execution of tests in controllers and supervisory systems, and validation of control algorithms

Contribuições a técnicas de agrupamento e visualização de dados multivariados utilizando mapas auto-organizáveis

Self-organizing maps (SOM) are artificial neural networks widely used in the data mining field, mainly because they constitute a dimensionality reduction technique given the fixed grid of neurons associated with the network. In order to properly the partition and visualize the SOM network, the various methods available in the literature must be applied in a post-processing stage, that consists of inferring, through its neurons, relevant characteristics of the data set. In general, such processing applied to the network neurons, instead of the entire database, reduces the computational costs due to vector quantization. This work proposes a post-processing of the SOM neurons in the input and output spaces, combining visualization techniques with algorithms based on gravitational forces and the search for the shortest path with the greatest reward. Such methods take into account the connection strength between neighbouring neurons and characteristics of pattern density and distances among neurons, both associated with the position that the neurons occupy in the data space after training the network. Thus, the goal consists of defining more clearly the arrangement of the clusters present in the data. Experiments were carried out so as to evaluate the proposed methods using various artificially generated data sets, as well as real world data sets. The results obtained were compared with those from a number of well-known methods existent in the literature

Comportamento de chamamento de Thyrinteina arnobia (Stoll) (Lepidoptera: Geometridae) em Psidium guajava, Eucalyptus grandis e em dieta artificial

Este estudo teve como objetivos determinar e comparar o efeito de três dietas: Psidium guajava, Eucalyptus grandis e dieta artificial no comportamento de chamamento e no padrão temporal do comportamento de chamamento de fêmeas virgens de Thyrinteina arnobia (Stoll). Foram analisados os parâmetros: tempo médio para o início do chamamento, número médio de chamamento, tempo médio de cada chamamento e tempo médio total de chamamento. Os bioensaios relacionados ao comportamento de chamamento de fêmeas virgens foram realizados durante seis escotofases consecutivas e as observações tomadas a cada 5 min., a 22 ± 1ºC, 70 ± 5% de UR e 10h de escotofase. As fêmeas, independente da dieta analisada, apresentaram duas posições diferentes de chamamento. A maioria das fêmeas virgens iniciou o chamamento na 1ª hora da 1ª escotofase. O padrão de chamamento da fêmea individualizada foi característico de um padrão contínuo. O tipo de dieta oferecida na fase larval influenciou o início do comportamento de chamamento (pré-chamamento) e o tempo médio total de chamamento. O tipo de dieta também alterou a coloração da glândula de feromônio.

Sites of Defoliation by Atta sexdens rubropilosa (Hymenoptera, Formicidae) in Artificial Plants

This study aimed to achieve a better understanding about the foraging behavior of leaf-cutter ant (Atta sexdens rubropilosa Forel) workers with respect to defoliation sites in plants. To accomplish that, artificial plants 70 cm in height were prepared and divided into four levels (heights), having natural plant leaves attached to them. Evaluations during the bioassays included the number of leaves dropped by the ants, as well as the percentage of plant mass removed. In all replicates, it became evident that the most exploited plant site is the apical region, which significantly differed from other plant levels.

Effectiveness of Photogem (R) activated by LED on the decontamination of artificial carious bovine dentin

The aim of this study was the evaluation of the effectiveness of photodynamic therapy on the decontamination of artificially induced carious bovine dentin, using Photoge(R) as the photosensitizer agent and an LED device as a light source. Dentin samples obtained from bovine incisors were immersed in sterile broth supplemented by Lactobacillus acidophillus 10(8) colony formation units (CFU) and Streptococcus mutans 10 8 CFU. Different concentrations of photosensitizer, PA = 1 mg/ml, PB = 2 mg/ml, and PC = 3 mg/ml, and two fluences, D = 24 J/cm(2) and D = 48 J/cm(2), were investigated. After CFU counting per milligram of carious dentin and statistical analysis, we observed that the photodynamic therapy (PDT) parameters used were effective for bacterial reduction in the in vitro model under study. The best result was achieved with the application of Photoge(R) at 2 mg/ml and photoactivated under 24 J/cm(2) showing a survival factor of 0.14. At higher photosensitizer concentrations, a higher dark toxicity was observed. We propose a simple mathematical expression for the determination of PDT parameters of photosensitizer concentration and light fluence for different survival factor values. Since LED devices are simpler and cheaper compared to laser systems, it would be interesting to verify their efficacy as a light source in photodynamic therapy for the decontamination of carious dentin.

Susceptibilidade de Rhodnius neglectus, Rhodnius robustus e Triatoma infestans (Hemiptera, Reduviidae, Triatominae) à infecção por duas cepas de Trypanosoma cruzi (Kinetoplastidae, Trypanosomatidae) utilizando xenodiagnóstico artificial

A susceptibilidade de ninfas de 3º estádio de Rhodnius neglectus, R. robustus e Triatoma infestans às cepas Y e AMJM de Trypanosoma cruzi foi verificada utilizando xenodiagnóstico artificial. Para a leitura do xenodiagnóstico, as fezes dos triatomíneos foram examinadas a cada dois dias, a partir do 5º até o 31º dia pós infecção, pela técnica de compressão abdominal. Os resultados mostraram diferenças na susceptibilidade dos triatomíneos para as duas cepas estudadas e o período ótimo de leitura variou do 11º ao 19º dias para a cepa Y e do 11º ao 15º dias para a cepa AMJM. Também, pôde-se concluir que para a cepa Y, as três espécies de triatomíneos demonstraram boa susceptibilidade, enquanto para a cepa AMJM, a melhor susceptibilidade foi observada com R. neglectus, seguida pelo T. infestans e R. robustus.

Modeling and analysis of artificial neural networks applied in operations research

Artificial neural networks are dynamic systems consisting of highly interconnected and parallel nonlinear processing elements. Systems based on artificial neural networks have high computational rates due to the use of a massive number of these computational elements. Neural networks with feedback connections provide a computing model capable of solving a rich class of optimization problems. In this paper, a modified Hopfield network is developed for solving problems related to operations research. The internal parameters of the network are obtained using the valid-subspace technique. Simulated examples are presented as an illustration of the proposed approach. Copyright (C) 2000 IFAC.

Artificial neural networks applied in study of atmospheric parameters to high voltage substations concerning lightning

This gaper demonstrates that artificial neural networks can be used effectively for estimation of parameters related to study of atmospheric conditions to high voltage substations design. Specifically, the neural networks are used to compute the variation of electrical field intensity and critical disruptive voltage in substations taking into account several atmospheric factors, such as pressure, temperature, humidity, so on. Examples of simulation of tests are presented to validate the proposed approach. The results that were obtained by experimental evidences and numerical simulations allowed the verification of the influence of the atmospheric conditions on design of substations concerning lightning.

Nitrogen content identification in crop plants using spectral reflectance and artificial neural networks

The accurate identification of the nitrogen content in crop plants is extremely important since it involves economic aspects and environmental impacts. Several experimental tests have been carried out to obtain characteristics and parameters associated with the health of plants and its growing. The nitrogen content identification involves a lot of nonlinear parametes and complexes mathematical models. This paper describes a novel approach for identification of nitrogen content thought spectral reflectance of plant leaves using artificial neural networks. The network acts as identifier of relationships among pH of soil, fertilizer treatment, spectral reflectance and nitrogen content in the plants. So, nitrogen content can be estimated and generalized from an input parameter set. This approach can be form the basis for development of an accurate real time nitrogen applicator.

Automatic detection of thermal damage in grinding process by artificial neural network

Esse trabalho tem por objetivo o desenvolvimento de um sistema inteligente para detecção da queima no processo de retificação tangencial plana através da utilização de uma rede neural perceptron multi camadas, treinada para generalizar o processo e, conseqüentemente, obter o limiar de queima. em geral, a ocorrência da queima no processo de retificação pode ser detectada pelos parâmetros DPO e FKS. Porém esses parâmetros não são eficientes nas condições de usinagem usadas nesse trabalho. Os sinais de emissão acústica e potência elétrica do motor de acionamento do rebolo são variáveis de entrada e a variável de saída é a ocorrência da queima. No trabalho experimental, foram empregados um tipo de aço (ABNT 1045 temperado) e um tipo de rebolo denominado TARGA, modelo ART 3TG80.3 NVHB.

Analysis of forecasting capabilities of ground surfaces valuation using artificial neural networks

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)