Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Fuzzy-basierte Leitkomponente zur multikriteriellen Optimierung von komplexen chemischen Prozessen

: Bernard, T.

Volltext urn:nbn:de:0011-n-244260 (93 KByte PDF)
MD5 Fingerprint: 9f65a673497beafc623b466253e1066f
Erstellt am: 04.02.2005

Sawodny, O. ; TU Ilmenau:
Synergies between information processing and automation. Vol.1 : 49. Internationales Wissenschaftliches Kolloquium, 27.-30.9.2004, conference proceedings
Aachen: Shaker, 2004
ISBN: 3-8322-2824-1
Internationales Wissenschaftliches Kolloquium <49, 2004, Ilmenau>
Konferenzbeitrag, Elektronische Publikation
Fraunhofer IITB ( IOSB) ()

In the automation of procedural processes in many cases problems of multi criteria optimization arise. Mostly this problems are solved by methods of Pareto Optimization, although this approach has the disadvantage of little transparency and high computational effort. Typical multi criteria optimization problems of procedural processes are the optimization of recipes of chemical processes or the choice of operating points of controlled processes in the area of steel and glass industry.
In all these applications several performance criteria have to be optimized instantaneously with different priorities. In many cases the performance criteria are fuzzy by nature. Other criteria may be well defined and good models may be available, e. g. economic criteria like material or energy losses. One suited approach to cope with well defined and fuzzy performance criteria is the concept of Fuzzy Decision Making by Bellman and Zadeh. The main advantages of this approach are its high level of transparency, the possibility of transparent weighting of single performance criteria and its low implementation effort.
In this paper a case study of the chemical industry is presented. The performance of the method of Fuzzy Decision Making is investigated and compared with the approach of Pareto optimization. The multi objective optimization of a non-linear chemical process with 4 input variables and 44 output variables is analyzed. It is shown that Fuzzy Decision Making yields higher performance indices, the results are much more transparent and the algorithm has much less computational effort. Furthermore it was demonstrated that a supplementary weighting of individual performance criteria is possible in a simple and transparent way. Thus, the optimization result can be adjusted iteratively to the needs of the user.