| Grant number: | 12/14928-2 |
| Support Opportunities: | Scholarships in Brazil - Doctorate |
| Effective date (Start): | December 01, 2012 |
| Effective date (End): | February 01, 2015 |
| Field of knowledge: | Engineering - Chemical Engineering |
| Principal Investigator: | Nivaldo Aparecido Corrêa |
| Grantee: | Elaine Schornobay |
| Host Institution: | Escola de Engenharia de São Carlos (EESC). Universidade de São Paulo (USP). São Carlos , SP, Brazil |
Abstract The modeling of air pollutants has been widely used as a tool for managing air quality. The prediction of particulate matter concentration may be more difficult than the prediction common pollutants gas, due to the greater complexity of the processes that control the formation, transport and removal of particles from the atmosphere. In recent years, some studies have been conducted with the use of artificial neural networks, in which prediction is a major application area. Unlike traditional models, the ANNs are self-adaptive methods, learning occurs through examples and capture subtle of functional relationships among the data, even those that are not clearly understood or are hidden. Thus, the ANNs are suitable for problems in which solutions require knowledge that it is difficult to specify but in which there are data or observations. Based on these characteristics, the objective of this work is to build two models to predict the concentration of particulate matter in Sao Carlos, as a tool using artificial neural networks. The models will be called RNA1 and RNA2. In RNA1, the forecast that will be held is monthly average concentration of particulate matter and RNA2 of the average concentration of next day of particulate matter, both will make the prediction of PM10 and PM2, 5. For training the neural network data will be used the concentration of particulate matter collected in the years 1997 to 2006, obtained by group of Environmental Control of UFSCar. A new stage of data collection will be performed to meet the needs of daily data at the same location and with the same equipment of previous data. Also will be used data climatic variables, provided by the National Institute of Meteorology (INMET). For use of weather data will be conducted analyzes of sensitivity, in order to identify the variables most relevant to the concentration of particulate matter. For the construction of the RNA1, both old data as new concentration will be turned into monthly averages. For the construction of RNA2, will be used only new data, which will feature daily values of collects. It is expected that the RNA1, proposed herein, is able to predict the concentration of particles without the use of the same previous concentrations. And that, RNA2 is a valuable tool for forecasting of the next day, moreover, that the modeling with neural networks for low and medium concentrations of particulate matter should be important for an understanding of the phenomena that surround them. Furthermore, the sensitivity analysis used in conjunction with neural networks will be an unprecedented contribution and can significantly reduce costs of monitoring and computer processing. | |
| News published in Agência FAPESP Newsletter about the scholarship: | |
| More itemsLess items | |
| TITULO | |
| Articles published in other media outlets ( ): | |
| More itemsLess items | |
| VEICULO: TITULO (DATA) | |
| VEICULO: TITULO (DATA) | |