Product quality monitoring using extreme learning machines and bat algorithms : a case study in second-generation ethanol production
ARTIGO
Inglês
In this study, a new methodology for online monitoring of second-generation ethanol production is presented. The prediction of the concentration of ethanol, substrate and cells from secondary measurements (pH, turbidity, CO2 and temperature) is compared with experimental data from the fermentation...
In this study, a new methodology for online monitoring of second-generation ethanol production is presented. The prediction of the concentration of ethanol, substrate and cells from secondary measurements (pH, turbidity, CO2 and temperature) is compared with experimental data from the fermentation of a mixture of molasses and hydrolyzed sugarcane bagasse from the alkaline hydrogen peroxide pre-treatment at 25 % and 75 % of volume. The Extreme Learning Machine algorithm (ELM) provided a very good alternative to traditional Multilayer Perceptron neural networks (MLP) and the BAT optimization technique applied to ELM algorithm provided a fast parallel search for the best solution. This new methodology offered a good alternative to the standard soft- sensor approach based on MLP and fast and reliable product quality estimates for key process variables as in second-generation ethanol production
Fechado
Product quality monitoring using extreme learning machines and bat algorithms : a case study in second-generation ethanol production
Product quality monitoring using extreme learning machines and bat algorithms : a case study in second-generation ethanol production
Fontes
Computer - aided chemical engineering Vol. 33 (2014), p. 955-960 |