Distribution-Free Approach to Probabilistic Model-Set Identification
- 303 Downloads
A model-set identification algorithm is proposed in a probabilistic framework based on a leave-one-out technique. It provides a nominal model and a bound of its uncertainty for a provided plant assuming that the effect of past inputs decays with a known bound. Because it does not require further assumptions on true plant dynamics or on noise, the risk of making inappropriate assumptions is small. The number of assumptions is shown to be minimum in the sense that identification is impossible after removing the assumption made here. Generalization of the proposed algorithm is considered in several aspects. A simple plant is identified for illustration.
KeywordsModel-set identification Leave-one-out estimation Linear programming Statistical learning theory Robust control
Unable to display preview. Download preview PDF.
- P. Khargonekar and A. Tikku, “Randomized algorithms for robust control analysis and synthesis have polynomial complexity,” inProceedings of the 35th IEEE Conference on Decision and ControlKobe, Japan, December 1996, pp. 3470–3475.Google Scholar
- M. Vidyasagar and R. L. Karandikar, “System identification: A learning theory approach,” inProceedings of the 40th IEEE Conference on Decision and ControlOrlando, FL, December 2001, pp. 2001–2006; also in this volume.Google Scholar
- M. Vidyasagar and R. L. Karandikar, “On the behaviour of learning algorithms with ß-mixing process inputs,” submitted for publication.Google Scholar
- E. Weyer and M. C. Campi, “Finite sample properties of system identification methods,” inProceedings of the 38th IEEE Conference on Decision and ControlPhoenix, AZ, December 1999, pp. 510–515.Google Scholar
- E. Weyer and M. C. Campi, “Non-asymptotic confidence ellipsoids for the least squares estimate,” inProceedings of the 39th IEEE Conference on Decision and ControlSydney, Australia, December 2000, pp. 2688–2693.Google Scholar