|
Poster Presentation 3-57
Simulation of an Aerated Lagoon Using Artificial Neural Networks and Multivariate Regression Techniques
Karla Patricia Oliveira-Esquerre1, Aline C. da Costa1, Milton Mori 1, Roy Edward Bruns2
1 Departamento de Processos Químicos, Faculdade de Engenharia Química, UNICAMP P.O. Box 6066, 13081-970 Campinas, SP - Brazil
2Instituto de Química, UNICAMP P. O. Box 6154, 13083-970 Campinas, SP - Brazil
Telephone: 55-19-3788-3963; Fax: 55-19-3788-3965; E-mail: karla@feq.unicamp.br
In recent years, computer-based methods have been applied to many areas of environmental chemistry. In the process industry the use of modern control strategies is required due to increasing demands on the quality of its effluent biochemical treatment systems.
The aim of this study is to develop an estimation model that can provide accurate predictions of the biochemical oxygen demand (BOD) of the output stream from the aerated lagoon at International Paper of Brazil, one of the major pulp and paper plants in Brazil.
Predictive models calculated from artificial neural networks (Functional Link Neural Networks - FLNN) and multivariate regression techniques (Multiple Linear Regression - MLR, Principal Components Regression - PCR and Partial Least Squares Regression – PLSR) are presented for BOD estimation.
The results show that neither artificial neural networks nor multivariate regression techniques are satisfactory when used separately in modeling and simulation. Best prediction performance is achieved when the data are preprocessed using Partial Least Squares (PLS) before they are fed to a FLNN. The PLS technique orthogonalizes the original input variables and helps FLNN nonlinear mapping. The influence of input variables is analyzed and satisfactory prediction results are obtained for an optimized situation.
|