Search

A robótica móvel é uma área em grande crescimento na pesquisa científica e na indústria, devido a sua versatilidade e capacidade de substituir os humanos, especialmente em trabalhos perigosos, e por apresentar melhor desempenho em certas aplicações. Várias delas são em ambientes externos, como em indústrias, aplicações aeroespaciais, petroquímicas, dentre outras. Esse artigo mostra a implementação de um sistema de localização de um robô móvel em ambiente outdoor, utilizando sistema de navegação global por satélite e odometria. Pretende-se, assim, obter uma forma robusta de localização, permitindo uma navegação confiável em diferentes terrenos. Testes foram realizados com a navegação por satélite e a odometria. Percebeu-se que o robô pôde trabalhar bem em ambientes externos capturando corretamente a localização GPS quando comparado com o Google Earth Pro. Com relação a odometria, houve pequenos erros quando o robô se moveu em linha reta, o que aumentou ao se realizar rotações. Isso se deve à característica do funcionamento do robô diferencial 4WD, onde as curvas geram deslizamento das rodas. Além disso, pode‑se demonstrar o bom funcionamento quando feita a fusão dos dados de GNSS, IMU e odometria.

Nos últimos anos, houve um crescimento na robótica móvel nas mais distintas áreas, impulsionado pela versatilidade de tarefas que esses robôs podem realizar. Entretanto, para que sejam capazes de executá-las de forma precisa, é essencial que possam navegar de maneira confiável, o que não é uma tarefa simples devido à dificuldade em se obter a pose exata do robô (posição e orientação). Este artigo descreve o desenvolvimento de um robô com a instrumentação necessária para testes de navegação autônoma e, para isso, usa encoders em cada par lateral de rodas para controlar a velocidade e calcular a odometria do robô. Para solucionar os erros acumulativos associados aos algoritmos de odometria, o robô também foi equipado com um sensor laser LiDAR para o uso futuro de algoritmos de localização, como o Monte Carlo Localization, e um sensor inercial para fusão de dados. O controle de dados e comandos é feito através do ROS e de um joystick e os resultados mostraram que o robô pôde controlar a velocidade e a posição de em um ambiente interno. Além disso, testes com o LiDAR para futuros algoritmos de localização absoluta foram realizados a fim de preparar o robô para implementações que exigem navegação autônoma e precisa.

Introduzem-se conceitos técnicos sobre câncer epitelial ou melanoma – um dos mais comuns e agressivos tipos de cancros existentes – e sobre o uso de aplicações ou ferramentas baseadas em técnicas de inteligência artificial e a influência da aplicação do aprendizado de máquina para realização de pré-diagnósticos na medicina em geral, que aceleram o tempo de descoberta da enfermidade e aumentam exponencialmente as chances de cura do enfermo acometido. Desenvolve-se o detalhamento técnico do algoritmo baseado em Deep Learning YOLOv5. Aborda-se a possibilidade de aplicação desta ferramenta para variados fins. Analisa-se o dataset utilizado nesta pesquisa e verificam-se as possibilidades de caracterização de tumores entre benignos e malignos, bem como classificam-se os tipos de formações epiteliais encontradas. Aborda-se a possibilidade de utilizar destas classificações como entrada para algoritmos baseados em aprendizado profundo por varredura de imagens. Conclui-se ser possível a criação de filtros lógicos que utilizem das leituras dos dados de saída providos pela execução do algoritmo YOLOv5 sob uma base fotográfica de exames dermatoscópicos, obtendo níveis de acerto relevantemente altos que apontariam com grande precisão a possível existência de malformações cancerígenas, facilitando com este o pré-diagnóstico e o combate precoce ao câncer de pele, aumentando consideravelmente a sobrevida de pacientes.

The generalized predictive controller (GPC) is an efficient strategy for controlling processes with time-varying parameters, as long as the GPC tuning parameters are chosen correctly. This study aims to present a new online tuning algorithm for the parameters of the GPC. The controllers are initially tuned by a model simulation (offline), via genetic algorithm, seeking quick answers and a small error. After variations in the setpoint, injection of disturbances in the output of the plant, and variations in the gains of the system operating in closed loop, the algorithm performs an online adjustment of these parameters using Fuzzy Logic. Based on the error information between the setpoint and the controlled variable and the variation of this error, the algorithm readjusts the tuning parameters of the GPC, so the performance of the control system response is not degraded. The algorithm is validated via model simulations representing the main characteristics of industrial plants. In the simulations, tests are presented by applying disturbances in the output of the plant, changing the dynamics of the model, and changing the setpoint. It is shown that the performance indexes of each plant are presented as being at least similar to those presented in [1], because it is still widely used in recent applications, and in some cases of variation of the dynamics of the plant, the proposed algorithm remained with a satisfactory result, while the presented by [1] became unstable.

No quesito navegação de robôs móveis em ambientes internos, um dos maiores problemas encontrados é a obtenção correta da localização do robô. Dentre os diferentes métodos existentes, há poucos que são satisfatórios para assegurar uma navegação correta por acumular erros, como o caso da dead reckoning ou emprego de sensores inerciais, ou pela limitação e imprecisão quando se trata de ambientes internos como no caso do GPS. O presente trabalho traz a implementação da fusão dos dados da dead reckoning e giroscópio para estimar a orientação de um robô móvel, utilizando o Filtro de Kalman Estendido (EKF) em uma plataforma robótica real, a fim de comparar o resultado da localização obtida com a fusão e com a dead reckoning. Para análise dos resultados, foi utilizado um sistema de localização por imagem, testado e validado, para fornecer a localização real do robô.

Due to the complexity and high financial costs involved in production processes, the steel industry can benefit from applications of intelligent systems, capable of performing automated activities. This research paper addresses a description of the process of creating a data-driven computational system to develop a computational thermal model of a real steel plate reheating furnace. Sufficiently accurate computational models can be used in conjunction with combustion control optimization techniques, such as model-based predictive control (MPC), or even a Digital Twin of the combustion system of a plate reheating furnace. The tool can be used in predictive failure diagnosis, fundamental for the maintenance and operation teams responsible for asset management. For this development, Recurrent Artificial Neural Networks have been widely applied, validating the existence of series that have temporal links between their samples, a typical case of monitoring industrial process variables. To meet the proposed objective, the performance of models based on recurrent neural networks of the Long Short Term-Memory (LSTM), Gated Recurrent Unit (GRU), and Temporal Convolutional Network (TCN) type was analyzed. The results were evaluated under different prediction horizons, since such techniques demand models capable of accurate predictions that are several steps ahead, premised on prediction capability.

Motores elétricos são onipresentes em processos industriais e no dia a dia e assim o diagnóstico de defeitos em tais máquinas é de suma importância. As técnicas mais recentes de diagnóstico utilizam CNN com grande sucesso, porém são implementadas para rodar em servidores off-site. Este trabalho analisa duas propostas de CNNs desenvolvidas para serem executadas em um sistema embarcado on-site. As CNNs foram testadas com sucesso por seus autores utilizando um conjunto de dados proprietário. Os testes realizados aqui utilizam um conjunto de dados diferente, muito utilizado em benchmarks na área, o conjunto de dados da CWRU. Os testes realizados demonstram que as CNNs propostas também obtêm sucesso ao classificar defeitos do conjunto de dados CWRU, validando suas estruturas e demonstrando uma boa generalização para ser aplicada em montagens e sistemas diferentes do utilizado em seu desenvolvimento.

This article presents the implementation of a specialist system with diffuse logic for controlling a fan in a mobile grid oven in an iron ore pelletizing plant. These fans are important to maintain the temperature of the pellets at predetermined values, this process is multivariable and has multiple restrictions, thus, diffuse logics were chosen for their ability to adapt to different operating conditions. The constraints and experiences of the employees were used to create rules with the objective of improving the efficiency of the process, in the article results are presented showing the improvement of the performance.

A pandemia do COVID-19 tem afetado a vida das pessoas bem como as economias de varios paises, os setores de saude, educacao, transporte, entre outros. Para tentar conter a disseminacao do virus, diversos paises implementaram barreiras sanitarias em aeroportos, rodoviarias, estacoes, portarias das empresas e outros espacos compartilhados para detectar pacientes com algum sintoma da infeccao viral. Visto que a febre e um dos sintomas mais recorrentes da doenca, iniciou-se uma corrida aos mercados por dispositivos de medicao de temperatura corporal. As câmeras termograficas, tambem conhecidas como termovisores, sao outros dispositivos utiliza-dos para medir a temperatura, empregando tecnologia conhe-cida como termografia infravermelha, uma ferramenta nao invasiva, rapida e objetiva. Neste estudo aplicouse a transferencia de aprendizado de maquina na YOLO para detectar as regioes mais quentes da face humana em imagens termograficas, permitindo a identificacao de estado febril em humanos. Para isso, os algoritmos de inteligencia artificial detectam as regioes de interesse nas imagens termograficas, que sao: os olhos, a testa e os ouvidos e em seguida, sao analisadas as temperaturas nestas regioes. O software desenvolvido apresentou excelente desempenho na deteccao das regioes de interesse estabelecidas, que indica adequadamente a maxima tempera-tura dentro das regioes de interesse, e que a escolha do metodo de maxima temperatura se apresentou adequada.

Keeping the level of steel in the mold of the continuous casting process constant is fundamental for the quality of the steel produced and, consequently, its commercial value. It is challenging, considering the several disturbances that cause undesired variations in the mold level. The aim of this paper is to apply a repetitive structure composed of two controllers, a generalized predictive controller (GPC) and a repetitive GPC (R-GPC) with constraints to mitigate the bulging and clogging/unclogging disturbances and the casting speed variation in the mold level of the process. The R-GPC controller has the same characteristics as the GPC, such as performance, robustness to disturbances, and insertion of constraints, and its advantage is the elimination of periodic disturbances. The repetitive structure will be implemented with a robustness filter and tuned by a genetic algorithm (GA). The controller tests are performed by simulations of a nonlinear mathematical model of the mold level, validated using real data from the steel industry. The proposed controller reduces the bulging disturbance amplitude by 98.5% and at 25% of the frequency of reversions in the valve. Consequently, the proposed controller allows an increase in the valve life span, a reduction in maintenance costs, and quality improvement in the steel slab.

In the steelmaking process by conversion of pig iron into an LD converter, it is necessary to inject oxygen through a metal lance. During this process, part of the slag accumulates in the body of the lance, forming the so-called skull. The presence of this skull on the copper nozzle of the lance or on its body is critical, because in the stage of its own use, there will be a heating and possible fusion of the skull, helping to make the fusion of the copper possible, causing in crevices where the water would flow into the converter, thus increasing the risk of explosion. The skull on the surface of the lance can also create a layer of enough volume to prevent the movement of the lance through the flanges located on the converters. Such occurrence and practice of removal impair operator safety, the environment and productivity. In this project, a system for measuring the thickness of the oxygen lance skull of LD converters by computational vision was developed, then the actions related to the maintenance of the lance can be taken in a timely manner, thus avoiding possible problems.

The world’s coffee agribusiness is worth US$ 91 billion annually and involves half a billion people. It is in this market that the interest of the Brazilian coffee production chain is centered, which contributed with more than 30% of the world production in the last harvests. The coffee market is characterized by a range of activities, complexity, dynamism, and a growing level of consumer demand for beverage quality. This imposes a high quality control on producer, consumer and exporter countries. Currently, the definition of quality and hence the value of coffee is based on manual grading, ie a person performs the role of a trained (certified) classifier to qualify coffee samples. Thus, the current process of classification of coffee suffers from the subjectivity of the classifiers and a great difficulty of standardization of the process due to possible inconsistencies. The present work proposes the use of an MLP (Multilayer Perceptron) Neural Network for analysis of coffee beans samples by digital image analysis, in order to increase the speed and reduce the subjectivities involved in the current manual classification process, considering: shape, size and color. Among the benefits of the automation of the coffee classification process are the reduction of costs, the agility and the standardization of the classification.

Mold level control is one of most important control loop in steel continuous casting process, because level variations higher than 6 - 10mm may impair the surface quality of the final products. The goal of this paper is to apply the Repetitive Generalized Predictive Controller (R-GPC) to control the mold level, considering bulging disturbances. Through simulations the optimum performance of this controller is shown when the model is affected by the bulging disturbances.

Electrical submersible pumps have been increasingly used in the oil industry. The startup cycle of this equipment causes it to be subjected to extreme conditions that can reduce its useful life. In addition, the oil flow produced during this process is lower than the normal operating condition. Currently, the control of this process is manual and subject to operator experience and sensitivity. In this way, mechanisms that allow to evaluate and increase the efficiency of this process are necessary. The methodology proposed in this work applies clustering technique to identify the possibility of reducing process time and was feasible.

Widely used in industries the traditional PID controller is without doubt the most far-reaching in industries. Although quite established, recent studies have shown a new concept for PID control, called fractional PID. Due to the nature of dynamic systems which mostly are best characterized by models of non-integer order differential equations, fractional PID is characterized by a more realistic reproduction of the nature of the system. In this paper it is presented and evaluated the performance of integer and fractional PIDs controllers tuned by genetic algorithm method applied to a level control plant. The results of simulations and practical tests are presented and through performance indicators integer and fractional PIDs controllers are compared.

This paper proposes a mathematical model of a process steam distribution network, with parameters set through a genetic algorithm, with the purpose of maximizing the energy utilization in a steel plant. The model is based on thermodynamics principles, including the balance of mass and energy. To determine the model parameters, due the dependencies between the equations, a genetic algorithm was applied. Once the control systems are not usually interconnected, steam pressure oscillations may occur in the Coke Plant side or in the Utilities Sector resulting in a large amount of wasted steam through independent relief systems. On a steel plant, steam can be used for some specific equipments or processes. In general, steam is considered a power source, applied on heating and power plants for electricity generation. Avoiding the steam waste is possible to improve the energy utilization figures resulting in both financial and environmental gains.

The Dynamic Matrix Control (DMC) Algorithm is a control method widely applied to industrial processes. Evolutionary Computation (EP) is a vibrant area of investigation, with some of the least widely known approaches being Genetic Algorithm (GA), Ant Colony Optimization (ACO) and Particle Swarm Optimization (PSO) all of which can be used in optimisation problem. This work make a comparative study of the effectiveness of the three methods to optimize the tuning parameters of the Dynamic Matrix Controller for SISO (single-input single-output) and MIMO (multi-input multi-output) linear dynamical systems with constraints.

The Dynamic Matrix Control (DMC) Algorithm is a control method widely applied to industrial processes. The goal of this work is to apply the Ant Colony Optimization (ACO) to optimize the tuning parameters of the Dynamic Matrix Controller for SISO (single-input single-output) and MIMO (multi-input multi-output) linear dynamical systems with constraints. A comparison is made between the computational method proposed here with the Genetic Algorithm, showing advantages and disadvantages of these two methods of tuning.

The goal of this paper is to apply the Generalized Predictive Controller (GPC) to control the mold level in a continuous casting process, considering the effects of the clogging disturb and the argon gas noise. Using a linear mathematical model of a real plant, it is shown by simulation the necessity of including the recursive identification and the Kalman filter in the control loop, in order to satisfy the maximum deviation and the process restrictions. The numerical results show the efficience of the GPC controller when compared with the classical PID controller.

A reliable monitoring of abnormal situations is still a challenge in industrial chemical processes. The earlier its detection and diagnosis, the greater the possibility of mitigating losses. An alternative approach to carry out these tasks is to make use of signal processing tools. In this direction, the hidden Markov model (HMM) method is used to monitor the accumulation of ash deposits along a convective heat transfer section of a boiler belonging to a Kraft pulping mill in Brazil. The identified model was able to provide the current state of the process working as a warning in case of a coming abnormal event.

PID optimal parameters selection have been extensively studied, in order to improve some strict performance requirements for complex systems. Ziegler-Nichols methods give estimated values for these parameters based on the system's transient response. Therefore, a fine tuning of these parameters is required to improve the system's behavior. In this work, genetic programming is used to optimize the three parameters Kp , Ti and Td , after been tuned by Ziegler-Nichols method, to control a high-order process, a large time delay plant and a highly non-minimum phase process. The results were compared to some other tuning methods, and showed to be promising.