Bayesian Gaussian Process Regression for Biomass Estimation in Batch Biotechnological Processes

FERNANDO A. DI SCIASCIO; ADRIANA N. AMICARELLI

Rio Gallegos - Santa Cruz - Argentina

Workshop; XII Reunión de Trabajo en Procesamiento de la Información y Control; 2007

Universidad Nacional de la Patagonia Austral

This paper proposes a biomass concentration estimator for a batch biotechnological process based on Bayesian regression with Gaussian Process. On the basis of experimental data, a two-stage boostrap technique has been developed for the estimator design. In the first stage, the biomass data set was augmented with virtual filtered measurements, and in the second stage, the biomass estimator design was completed. The method provides information on the confidence level of the estimates, and the biomass estimator performances are illustrated for the bacillus thuringiensis delta - endotoxins production process.