A fast model identification method for networked control system

Most control strategies for networked control system (NCS) assume that system model is known as ‘a priori’, which are however impractical in many industrial applications. To obtain a suitable model for networked control, this paper concentrates on model identification under the networked control env...

Full description

Saved in:
Bibliographic Details
Published in:Applied mathematics and computation Vol. 205; no. 2; pp. 658 - 667
Main Authors: Fei, Minrui, Du, Dajun, Li, Kang
Format: Journal Article Conference Proceeding
Language:English
Published: Amsterdam Elsevier Inc 15.11.2008
Elsevier
Subjects:
Approximations and expansions
Cubic spline
Exact sciences and technology
Fast recursive algorithm
Mathematical analysis
Mathematics
Network-induced delay
Networked control system
Numerical analysis
Numerical analysis. Scientific computation
Numerical approximation
Sciences and techniques of general use
Networked control system
Network-induced delay
Fast recursive algorithm
Cubic spline
Numerical approximation
Interpolation
ARMA model
Numerical analysis
Applied mathematics
Control system
Fast algorithm
Recursive algorithm
ISSN:0096-3003, 1873-5649
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Most control strategies for networked control system (NCS) assume that system model is known as ‘a priori’, which are however impractical in many industrial applications. To obtain a suitable model for networked control, this paper concentrates on model identification under the networked control environment. A networked identification scheme is proposed here, and nondeterministic factors such as network-induced delay, data packet out-of-order, and data packet loss between the sensor and identifier as well as the identifier and the actuator are considered. A discard-packet strategy is first developed to deal with network nondeterministic factors from the sensor to the identifier, and a cubic spline interpolation is used to compensate lost data. Actuator buffers are then introduced for actuator nodes to handle network nondeterministic factors between the identifier and the actuator. Finally, a fast recursive algorithm is used both to select the model structure and to estimate the unknown system parameters. Experiments on ARMA model identification and NARMAX model identification are performed respectively under different network loads. Simulation results show that the proposed networked identification scheme can effectively overcome the impact of various network-induced nondeterministic factors and significantly improve model identification performance.
ISSN:0096-3003
1873-5649
DOI:10.1016/j.amc.2008.05.040