Nichtlineare Dynamik in der Chemischen Technik. Teilprojekt C7: Nichtlineare praediktive Regelung bei unsicheren Prozessmodellen Schlussbericht

In this work Nonlinear Model Predictive Control (NMPC) of batch- and semibatch reactors has been examined. Here, typical industrial scenarios were treated assuming limited process knowledge and measurement information. It has been shown that NMPC schemes in combination with Receding Horizon Estimation (RHE) can lead to considerable improvements even if only calorimetric models are available. If the feedflow rates are also determined by the controller on-line optimization can further increase the performance using a cost function which reflects the overall economics. In the second part of the project off-line optimization on the basis of a detailed model has been treated. Here, new methodologies for the incorporation of safety related failures and controllability issues have been developed. In cooperation with Bayer AG some results have been applied to an industrial polymerization reactor in laboratory and production scale. (orig.)SIGLEAvailable from TIB Hannover: F98B1258 / FIZ - Fachinformationszzentrum Karlsruhe / TIB - Technische InformationsbibliothekBundesministerium fuer Bildung, Wissenschaft, Forschung und Technologie, Bonn (Germany)DEGerman