Web projektu DAR

Research Centre
Data - Algorithms - Decision Making

Established in 2005 under support of MŠMT ČR (project 1M0572)

Publications

Smoothing in Multiple Model Change Detection for Stochastic Systems

Typ:

Conference paper

Authors:

Punčochář I., Duník J., Šimandl M.

Proceedings name:

Proceedings of the 8th ACD 2010 European Workshop on Advanced Control and Diagnosis

Serie:

Ferrara, Italy

Year:

2010

Pages:

79-84

Contact:

Ing. Ivo Punčochář, Ph.D. (FAV Západočeská univerzita, katedra kybernetiky)

Keywords:

Stochastic systems, detection systems, change detection

URL (www page):

http://www.ing.unife.it/eventi/ACD2010/

attachment1:

ADC10_IP.pdf [file]

attachment2:

prezentace_ADC10_IP.pdf [file]

Anotation:

The paper focuses on the application of smoothing in multiple model change detection for stochastic systems. In a typical multiple model change detection scheme, the decisions are made based on ?ltering estimates of an unmeasured state of an observed system. Since better estimates of the state lead to ?ner decisions, a higher quality of estimates is of great interest. The way to improve the decisions considered in this paper consists in deferring decisions and using more precise smoothing estimates instead of the ?ltering ones. As a result, the decisions of a higher quality are obtained at the cost of delaying that decisions. The approach introduces a new level of the compromise between the quality of decisions and the delay for detection that is inherent in all change detection methods.