Abstract
Control systems use sensors in the monitoring and control of process state. Sometimes, process variables are estimated because proper sensors are unavailable, they are prohibitively expensive or measurements are difficult to perform. One Solution consists in to infer the state variables which are not measured from other variables by using virtual sensors or software sensors (soft-sensors). In alcoholic fermentation processes, measuring the ethanol concentration is essential. However, there arc no cheap and reliable sensors capable of providing on-line measurements nor is there a global model for this variable which is accepted by everyone. In addition, two fermentations never are equal because microorganisms are very sensitive to small variables deviations. Therefore, these processes require an adaptive and robust estimating system. This paper presents an adaptive soft-sensor for a bioethanol fermentative process using all evolving fuzzy model. In addition, the obtained model is compact and it presents a Suitable structure for future application in control strategies, in order to optimize the process productivity and to reduce production Costs. Copyright (C) 2009 CEA.