"Successful modeling and identification are necessary prerequisites to good prediction, control and signal processing. Many of the processes and plants met in practice exhibit a nonstationary behavior. Also, process and plant models often describe the actual behavior only locally, around the current working point. Whatever the cause, reliable algorithms for tracking of slow parameter variations and detection of abrupt parameter changes are needed in most applications. The current state of the art oscillates between the theoretically optimal but rarely tractable concept of nonlinear filter designed for a particular model of parameter variations, and more or less heuristic, ad hoc techniques of forgetting and detection. The objective of the proposed project is to design and elaborate methods that would be feasible even with limited computer resources, yet close in a well-defined sense to the theoretically optimal solution. The emphasized topics will be identification of harmonic signals, advanced forgett"@en . . "0"^^ . "Identifikace nestacion\u00E1rn\u00EDch syst\u00E9m\u016F" . . . "1"^^ . . "Rozs\u00E1hl\u00E1 p\u016Fvodn\u00ED systematick\u00E1 pr\u00E1ce v oblasti identifikace nestacion\u00E1rn\u00EDch proces\u016F. \u0158e\u0161itelsk\u00FD t\u00FDm efektivn\u011B spolupracoval se zahrani\u010Dn\u00EDmi v\u011Bdci a pracovi\u0161ti. Podstatn\u00E1 v\u011Bt\u0161ina c\u00EDl\u016F a z\u00E1m\u011Br\u016F projektu byla spln\u011Bna. V\u00FDstupy projektu byly vyu\u017Eity pro doktor"@cs . "Identification of nonstationary systems"@en . . . "37"^^ . . "GA102/97/0466" . "37"^^ . . . . "http://www.isvav.cz/projectDetail.do?rowId=GA102/97/0466"^^ . . . . "0"^^ . "Dobr\u00E9 modelov\u00E1n\u00ED a identifikace jsou nezbytn\u00FDmi p\u0159edpoklady pro \u00FAsp\u011B\u0161n\u00E9 zpracov\u00E1n\u00ED sign\u00E1lu, predikci a \u0159\u00EDzen\u00ED. Mnoho proces\u016F, se kter\u00FDmi se setk\u00E1v\u00E1me v praxi, vykazuje nestacion\u00E1rn\u00ED chov\u00E1n\u00ED. To vyvol\u00E1v\u00E1 pot\u0159ebu spolehliv\u00FDch algoritm\u016F pro sledov\u00E1n\u00ED pomal\u00FDch a detekci n\u00E1hl\u00FDch zm\u011Bn parametr\u016F. V sou\u010Dasnosti pou\u017E\u00EDvan\u00E9 metody sledov\u00E1n\u00ED a detekce kol\u00EDsaj\u00ED mezi teoreticky optim\u00E1ln\u00ED, ale prakticky t\u011B\u017Eko pou\u017Eitelnou koncepc\u00ED neline\u00E1rn\u00EDho filtru, navr\u017Een\u00E9ho pro konkr\u00E9tn\u00ED model v\u00FDvoje parametr\u016F, a v\u00EDcem\u00E9n\u011B heuristick\u00FDmi technikami zapom\u00EDn\u00E1n\u00ED a detekce. C\u00EDlem navrhovan\u00E9ho grantu je navr\u017Een\u00ED a anal\u00FDza algoritm\u016F a metod, kter\u00E9 by byly vy\u010D\u00EDsliteln\u00E9 p\u0159i dan\u00FDch omezen\u00EDch na v\u00FDpo\u010Detn\u00ED n\u00E1ro\u010Dnost a kter\u00E9 by p\u0159itom poskytovaly \u0159e\u0161en\u00ED v dob\u0159e definovan\u00E9m smyslu bl\u00EDzk\u00E9 optim\u00E1ln\u00EDmu. Grant je zam\u011B\u0159en na identifikace parametr\u016F sign\u00E1lu, modern\u00ED strategie zapom\u00EDn\u00E1n\u00ED, modelov\u00E1n\u00ED a detekci zm\u011Bn parametr\u016F pomoc\u00ED strom\u016F v\u00FDvoje hypot\u00E9z, p\u0159ibli\u017En\u00E9 \u0159e\u0161en\u00ED \u00FAlohy neline\u00E1rn\u00ED filtrace a aplikac\u00EDm na vybran\u00E9 technologick\u00E9 a ekonomick\u00E9 proces" . .