Modeling of Nonlinear Processes using Wiener-NN Representation

• Published in: 2014 IEEE Conference on Norbert Wiener in the 21st Century (21CW) pp. 1 - 8
• Main Author: Visala, Arto
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-06-2014
• Subjects: chromatographic separation process; fermentation processes; identification; intelligent control; Kalman filter; Laguerre approximations; monitoring; nonlinear models; recurrent neural nets; robust identification; state-space models; supervision; Wiener model
• DOI: 10.1109/21CW31760.2014.10015719

A modeling method for nonlinear MIMO systems is introduced 'breeding' the best capabilities of the classical Wiener representation and NN-type models. The original Wiener representation consists of linear dynamics (Laguerre filters) and static nonlinear mapping (polynomial expansion). In Wiener-NN, static nonlinear mapping is approximated with NN. The feedforward Wiener-NN is suitable for modeling of finite settling time systems. The Wiener-NN with feedback can also be used for modeling of autonomous type systems. MLP has been used as NN. The Extended Kalman filter is used in the recursive estimation of the parameters of NOE Wiener-MLP with feedback. Biotechnical processes and chromatographic separation process are used as modeling cases.