Kalman Filter and Neural Network for On-Line Estimation of Polymer Chain Characteristics

Autor: C. Vidal, S. Guiochon, G. Defaye, L. Caralp
Rok vydání: 1995
Předmět:
Zdroj: IFAC Proceedings Volumes. 28:287-292
ISSN: 1474-6670
DOI: 10.1016/s1474-6670(17)47051-0
Popis: In the past, many efforts have been paid to provide a general way for on-line monitoring of various molecular characteristics while the polymerization reaction proceeds (e.g. average chain length, chain length distribution, etc.). An efficient tool for reaching such characteristics is supplied by the Kalman filtering. The main limitation to its use however, comes from the need of a relevant mathematical model of the process. It was decided to substitute a neural network-based model for the more common kinetic-based model of the average chain lengths evolution.
Databáze: OpenAIRE