Monday, May 4, 2009
9-31
An intelligent system for multivariate on-line monitoring of a continuous flash fermentation process
Elmer Ccopa Rivera1, Daniel Ibraim Pires Atala2, Aline Carvalho da Costa3, and Rubens Maciel Filho2. (1) School of Chemical Engineering, State University of Campinas, Caixa Postal 6066, Campinas, 13083970, Brazil, (2) Department of Chemical Processes, University of Campinas-UNICAMP, Albert Einstein Avenue, 500, Campinas, 13081970, Brazil, (3) Department of Chemical Processes, State University of Campinas, School of Chemical Engineering, State University of Campinas, P.O. Box 6066, Campinas, 13083-970, Brazil
This study presents results from the implementation and testing of a PC based monitoring system for a bioethanol production process using Artificial Neural Network-based software sensors (ANN-SS). The system is based on an array of primary sensor, a communication module and a monitoring and data acquisition subsystem. This integrated framework provides a real-time monitoring solution, which is one of the most important aspects of the decision making in the strategies of optimization and control of bioprocesses. A LabVIEW data acquisition module is implemented to read all influencing variables, which are first used to train the neural networks. The optimal ANN architecture is placed in a LabVIEW based program formula node that monitors the concentration of bioethanol. The ANN-based monitoring system represents thus a robust model-based approach which is expected to contribute for improving the implementation of suitable operating strategies of optimization as well as advanced control to achieve high operational performance.