(83e) Development of A Hybrid Neural System for Monitoring the Bioethanol Production with Yeast Recycling

Researches are being conducted worldwide for the development of new technologies for producing liquid fuels from renewable sources. In this context, Brazil, as a bioethanol producer from sugar cane aims, with this alternative, to reduce the crude petroleum consumption as well as the environment pollution. Within the proposed scenario, the goal is to contribute with proceedings and a computer operation technique that will help reducing production costs of ethanol biofuel

In this present work a software sensor system was developed to obtain the biomass substrate and product concentration rates from secondary measurements as: pH, turbidity, CO₂ rate and temperature. The sensor software uses a neural hybrid model to combine a multi;layer percepetion artificial neural network and the mass balance that describes the fermentation kinetic process. Experimental data used for training the hybrid model were obtained from fermentation performed at 30-38 º C, with yeast already used in other fermentation process (recycle cells), with a step of 2 ° C. Fermentation raw material is a is a mixture of hydrolyzed bagasse and cane molasses 75% and 25% volume, respectively. This type of composition is typical of the second generation ethanol production process coupled with the first generation process. The integrated structure allows making the monitoring of bioethanol production in real time.