Determinação de parâmetros cinéticos da gaseificação de biomassas do Sul do Brasil por meio de redes neurais artificiais

Abstract

This study applied artificial neural networks to establish the correlation between biomass constituents, ultimate analysis and proximate analysis, reactor specifications and the kinetic parameters (activation energy and pre-exponential factor) of biomass gasification. Three ANN models were developed, one for each of the two kinetic parameters and one for all parameters in a single ANN. The dataset included 76 experiments, comprising 40 data points from various biomasses found in the literature and 36 data points from widely used biomasses in the Southern region of Brazil. These data were used to train and test the neural networks. The study demonstrated effectiveness in predicting activation energy, both globally and specifically for biomasses from the Southern region of Brazil, achieving determination coefficients of 0.8571 and 0.9545, respectively. However, the artificial neural networks used for predicting the pre-exponential factor proved ineffective, with the lack of many values for this parameter in the dataset being the main reason for this.