Genetic programming applied to chemical and environmental engineering

• Tipo: Capítulo ou Parte de Livro
• Data: 2015

Autores

Afonso, NF (Autor) Outra; Martins, FG (Autor) FEUP; Pires, JCM (Autor) FEUP

Livro

Título: Horizons in Earth Science Research

Páginas: 191-207

ISBN: 978-163482581-8;978-163482550-4

Indexação

Scopus - 0 Citações

Outras Informações

ID Authenticus: P-00K-6ZD

Abstract (EN): Genetic programming (GP) is an evolutionary procedure that automatic generates models, optimizing simultaneously its structure and parameters. It is able to define linear or non-linear models based on input-output training datasets, using the same principles of the Darwinian Theory of Evolution. In this modelling procedure, detailed information of process phenomena is not required. GP can be used to a wide range of modelling applications, including Chemical and Environmental Engineering. In this study, two case studies were presented: (i) estimation of binary gas diffusion coefficients; and (ii) prediction of the next day hourly average ozone concentrations (O3, considered one of most concerning air pollutants in Europe). In the first case, GP was compared with artificial neural networks (ANNs) and other models obtained in literature. The determined ANN model presented as main disadvantages the high complexity (typical for this type of statistical model) and high number of parameters. In training set, the GP model presented one of the best performances. On the other hand, in test set, ANN predictions were the closest to experimental values. Considering the disadvantages of ANN model, the performance GP model showed that can be considered an alternative to the best models presented in the literature. In the second case, GP was able to determine the main (meteorological and environmental) variables that influences the O3 concentrations: temperature, relative humidity, NO2 and O3 concentrations of the previous day. Additionally, GP models achieved good predictive performance. © 2015 Nova Science Publishers, Inc.

Idioma: Inglês

Tipo (Avaliação Docente): Científica