Representing and Learning Unmodeled Dynamics with Neural Network Memories

A nonlinear model representation consisting of an interpolation of several local models, which are valid within certain operation regimes, is proposed. Using this representation, first principles models and black-box models like neural networks may be integrated. Only operation regimes of the plant...

Full description

Saved in:
Bibliographic Details
Main Authors: Johansen, Tor A., Foss, Bjarne A.
Format: Conference Proceeding
Language:English
Subjects:
Cybernetics
Data engineering
Economic forecasting
Equations
Mathematical model
Neural networks
Nonlinear control systems
Optimal control
Reliability engineering
Working environment noise
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:A nonlinear model representation consisting of an interpolation of several local models, which are valid within certain operation regimes, is proposed. Using this representation, first principles models and black-box models like neural networks may be integrated. Only operation regimes of the plant not adequately modeled by first principles are being represented and learned by a neural network memory. The principle is illustrated by simulation examples.
DOI:10.23919/ACC.1992.4792705