Identification of Wiener-MLP with feedback NOE-model with extended Kalman filter

The classical input-output Wiener-representation consists of linear dynamics (Laguerre filters) and static nonlinear static polynomial mapping. In Wiener-MLP the static nonlinear mapping is realized with MLP. By feeding back some of the outputs of Wiener-MLP, a model capable of modeling autonomous s...

Full description

Saved in:
Bibliographic Details
Main Author: Visala, A.
Format: Conference Proceeding
Language:English
Subjects:
Context modeling
Convergence
Convolution
Equations
Feedback loop
Kernel
Neural networks
Nonlinear dynamical systems
Predictive models
Recursive estimation
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The classical input-output Wiener-representation consists of linear dynamics (Laguerre filters) and static nonlinear static polynomial mapping. In Wiener-MLP the static nonlinear mapping is realized with MLP. By feeding back some of the outputs of Wiener-MLP, a model capable of modeling autonomous systems, like batch processes, can be realized. The dynamics contains MLP and Laguerre system in the feedback loop. The model can be presented in state-space form. The MLP can be interpreted as the measurement equation of the system. An extended Kalman filter is used in recursive NOE-type estimation of the MLP parameters in order to identify a model suitable for simulation. Wiener-MLP with feedback models are identified for two bioprocesses.
ISSN:1098-7576
1558-3902
DOI:10.1109/IJCNN.1998.685959