APPLICATION OF BAYESIAN TECHNIQUE FOR PARAMETER ESTIMATION IN A FERMENTATIVE PROCESS

In the current world scenario, there has been noted an increase of researches on biofuel production, more specifically bioethanol, produced from biomass, in order to obtain more information to analyze, understand and optimize this fermentative process. The modelling process, which include the determination of a kinetic model and its respective parameters, is a fundamental step in defining operating strategies and understand how the experimental conditions can affect the optimal system operating conditions. The present work employs a bayesian technique to estimate the parameters of a classical kinectic model used by Silva and collaborators (2016), because, unlike the classical techniques, it is possible to take into account the uncertainty of the measurements and the prior knowledge of the parameters can be accounted for in probabilistic terms. In this context, by using simulated measurements, for the parameters estimation it is propose a sensitivity analysis of the parameters model to define the most relevant ones to be estimate and the use of the Monte Carlo Markov Chain method through the Metropolis-Hastings algorithm, evaluating the influence of four types of priori probability distribution of data set: uniform, gaussian, log-normal and Rayleigh. The obtained results showed that the sensibility analysis is an important step on parameter estimation and algorithm used was satisfactory in estimating the parameters of the kinectic model used, demonstrating the possibility of using it as a tool for time and cost reduction in experimental tests.