Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Early detection and identification of undesirable states in chemical plants using neural networks

: Neumann, J.; Deerberg, G.; Schlüter, S.; Schmitt, W.; Hessel, G.

Keil, F.; Mackens, W.; Voss, H.; Werther, J.:
Scientific Computing in Chemical Engineering II. Vol.2: Simulation, Image Processing, Optimization and Control
Berlin: Springer, 1999
ISBN: 3-540-65851-3
Workshop on Scientific Computing in Chemical Engineering <2, 1999, Hamburg>
Fraunhofer UMSICHT Oberhausen ()
Fehlerdiagnose; Früherkennung; neuronales Netzwerk; Sicherheitstechnik; exotherme Reaktion; fault diagnosis; early detection; neural network; safety technology; exothermic reaction

The suitability of pattern recognition for safety diagnosis of chemical plants is discussed. Experiments in a miniplant and with a process simulator are carried out. The process characteristics are treated with different recognition methods and classified with the aid of expert know how. Afterwards, the trained system can be used for process diagnosis. The capability of neural networks for this problem can be shown.