Identificação e comparação das taxas de acerto e erro de transições de radiação por meio de dados medidos por um sensor de luminosidade arduino e um pireliômetro

• Ano de defesa: 2017
• Autor(a) principal: Campos, Lucas Freitas
• Orientador(a): Rocha, Paulo Alexandre Costa
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Engenharia mecânica; Energia solar; Radiação solar - Medição
• Link de acesso: http://www.repositorio.ufc.br/handle/riufc/24539

Energy is a very important natural resource for all living creatures. The research of this work is related to solar energy. In order to achieve a proper way to take advantage of this solar resource, it is necessary to analyze the places and times where solar radiation is highest. Within this context, clouds play an important role in the use of the solar energy resource, as they are responsible for covering the sun, thus reducing the direct solar radiation available at certain periods of the day. In this work a prototype developed in the LESGN of the UFC was applied. This device is called Solar Irradiation Time Meter (SITM), and uses an Arduino Micro controller board. The measurement is done by means of six LDR's (Light Dependent Resistor), generating data with the values of its resistances according to the luminosity emitted by the sunlight. Direct radiation data obtained with a pyrheliometer were also used. This work aims to evaluate these data obtained by these two devices and search through them to identify the presence of a cloud covering the sky. The arrangement of the generated data is done by a classifier that determines if at any given moment the data belongs to the class cloud or not cloud. The data is based on moving averages and the increments of them. From the classification results obtained by the data in the two sensors, the process of obtaining the confusion matrix of the same was done to make a correct comparison of the fidelity and error rates of the classification algorithm and to calculate the Kappa coefficient, which estimates the classification performance levels.