Parameter Estimation and Prediction of a Nonlinear Storage Model: an algebraic approach

T.G. Doeswijk, K.J. Keesman

Research output: Contribution to conferenceConference paperAcademicpeer-review

1 Citation (Scopus)

Abstract

Generally, parameters that are nonlinear in system models are estimated by nonlinear least-squares optimization algorithms. In this paper, if a nonlinear discrete-time model with a polynomial quotient structure in input, output, and parameters, a method is proposed to re-parameterize the model such that the model becomes linear in its new parameters. The new parameters can then be estimated by ordinary least squares. Finally, the model is rewritten in predictor form. A model of an agricultural storage facility with real data is presented to demonstrate the procedure and show the improved predictive performance. Some technical problems are indicated and solutions are proposed.
Original languageEnglish
Publication statusPublished - 2005
Event2005 International Conference on Control and Automation -
Duration: 27 Jun 200529 Jun 2005

Conference

Conference2005 International Conference on Control and Automation
Period27/06/0529/06/05

    Fingerprint

Cite this

Doeswijk, T. G., & Keesman, K. J. (2005). Parameter Estimation and Prediction of a Nonlinear Storage Model: an algebraic approach. Paper presented at 2005 International Conference on Control and Automation, .