Estimativa de concentração de material particulado em suspensão na atmosfera por meio da modelagem de redes neurais artificiais

• Ano de defesa: 2006
• Autor(a) principal: Monica Marques Caetano de Lima
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: meio ambiente e recursos hídricos; Saneamento; Meio ambiente; Ar Poluição; Redes neurais (Computação); Recursos hídricos Desenvolvimento; Engenharia sanitária
• Link de acesso: http://hdl.handle.net/1843/ENGD-6XXNAQ

The aim of this study was to predict the total particulate matter concentration in the main areas of Ipatinga region. The artificial neural networks (ANN) was the modelling tool used. This model is capable of predicting the pollutant concentration just by training the input and output parameters. The input parameters were meteorological such as wind direction, wind speed, rain, ambient temperature and temporal such as, summer and winter. The output parameter used was the historical data of the total particulate matter concentration taken between 1996 and 2004. In the modelling, the multilayer perceptron (MLP) model was tested. Among the MLP configurations evaluated, the topology 13-7-6 was chosen. The validation of the model was done by comparing the simulated with the observed values. This model was also compared with the industrial source complex short-term dispersion model (ISCST3). Thefour statistical tools used to evaluate the fitting were medium squared error (MSE), fractional bias (FB), index of agreement (IA) and linear correlation coefficient (R). Comparing the results it was seen that the predicted values were better in das Águas, Cariru and Bom Retiroboroughs and were overestimated in Novo Cruzeiro, Castelo and Ferroviários. Besides, the predicted results of the ANN model were better than the ISCST3 dispersion model.