On the Consistency of Prediction Error Identification Methods
1976 (English)In: System identification: advances and case studies / [ed] Raman K. Mehra, Dimitri G. Laintotis, New York, 1976, 121-164 p.Chapter in book (Refereed)
The problem of identification is to determine a model that describes input–output data obtained from a certain system. In this chapter, strong consistency for general prediction error methods, including the maximum-likelihood (ML) method is considered. The results are valid for general process models: linear and nonlinear. An error identification method is discussed in the chapter along with a general model for stochastic dynamic systems. Different identifiability concepts are also introduced, where a procedure to prove consistency is outlined. Consistency is shown for a general system structure, as well as for linear systems. The application of the results to linear time-invariant systems is also discussed in the chapter.
Place, publisher, year, edition, pages
New York, 1976. 121-164 p.
, Mathematics in Science and Engineering, ISSN 0076-5392 ; 126
Input-output data, Maximum likelihood, Error identification, Consistency
IdentifiersURN: urn:nbn:se:liu:diva-100811DOI: 10.1016/S0076-5392(08)60871-1ISBN: 0-12-487950-0OAI: oai:DiVA.org:liu-100811DiVA: diva2:663749