Gehlen C.,GIMOCAP PPGEPRO UFSM |
Koch G.G.,GIMOCAP PPGEPRO UFSM |
Franchi C.M.,GIMOCAP PPGEPRO UFSM |
Hoffmann R.,GIMOCAP PPGEPRO UFSM |
Salau N.P.G.,GIMOCAP PPGEPRO UFSM
Computer Aided Chemical Engineering | Year: 2012
The main indicator of the distilled ethanol quality is its composition. In general, the online composition analyzers are not available due to their high cost. To overcome the lacking of devices to measure online and at real-time the process composition, we have proposed a soft-sensor to work as a virtual analyzer. Among the techniques available in the literature to achieve this goal, we have chosen the identification and neural networks. Both are used to infer online the ethanol composition through the real-time measured temperatures. According to our results, the neural networks have shown better performance in the composition inference. Further, this technique as soft-sensor was implemented in a SCADA (Supervisory Control And Data Acquisition) software in order to monitor the distilled ethanol composition. © 2012 Elsevier B.V.