Practical Issues of System Identification
2007 (English)Report (Other academic)
System Identification concerns the problem of building mathematical models of dynamical systems. This involves a fair amount of theory and algorithms. Equally important, though, is the practical side of the methodology. This chapter deals with the issues that are essential to construct a good model in practice. Such issues include the problem of input and experiment design. Typical essential features of the input are discussed as well as examples of commonly used inputs. Next comes the question of how to condition the measured signals. This involves issues of removing trends and disturbances outside of the frequency ranges of interest for the model. The most demanding task is to find a suitable model structure, guided by information in the observed data. A first cut methodology for this is described. For a successful application, it is in the end a matter of combining intuition, and information from various data test. Two applications, a fighter aircraft and a buffer vessel in process industry illustrate the process.
Place, publisher, year, edition, pages
Linköping: Linköping University Electronic Press, 2007. , 25 p.
LiTH-ISY-R, ISSN 1400-3902 ; 2810
Parameter estimation, Model structure, Input design, Drift, Trends, Identification applications, PRBS, Chirp signal, ARX models, State-space models, Output-error models
IdentifiersURN: urn:nbn:se:liu:diva-56133ISRN: LiTH-ISY-R-2810OAI: oai:DiVA.org:liu-56133DiVA: diva2:316969