The smooth variable‐structure filter is a model‐based state estimation algorithm, which can be applied to linear and smooth nonlinear dynamic systems. In this algorithm, the source of uncertainty can be explicitly defined. Furthermore, convergence of the algorithm can be guaranteed, given an upper bound on the level of noise and uncertainties. Performance of the algorithm can be improved through refining the upper bounds on parameter variations or uncertainties. The filter innovation vector or (output) estimation error can be used as a measure of performance for estimation algorithms. The smooth variable‐structure filter benefits from using a secondary set of performance indicators in addition to the innovation vector. Such indicators reflect the effect of the corresponding modeling errors on each estimated state or parameter. Aiming at robustness against uncertainties, the internal model used in the filtering algorithm can be dynamically refined based on the multiple performance indicators. In fault diagnosis and prognosis, the two features of dynamic model refinement and robustness against uncertainty are of critical importance. Furthermore, the smooth variable‐structure filter can be combined with different Bayesian filters in order to achieve a trade‐off between robustness and optimality. The reviewed applications of the smooth variable‐structure filter include multiple target tracking, battery state‐of‐charge estimation, and robotics.


    Zugriff

    Zugriff prüfen

    Verfügbarkeit in meiner Bibliothek prüfen

    Bestellung bei Subito €


    Exportieren, teilen und zitieren



    Titel :

    Smooth Variable‐Structure Filter


    Beteiligte:
    Setoodeh, Peyman (Autor:in) / Habibi, Saeid (Autor:in) / Haykin, Simon (Autor:in)

    Erschienen in:

    Erscheinungsdatum :

    12.04.2022


    Format / Umfang :

    28 pages




    Medientyp :

    Aufsatz/Kapitel (Buch)


    Format :

    Elektronische Ressource


    Sprache :

    Englisch