Тип публикации: доклад, тезисы доклада, статья из сборника материалов конференций
Конференция: Aspire to Science; Новосибирск; Новосибирск
Год издания: 2018
Ключевые слова: Duhamel integral, transitional function, weight function, delta-shaped input
Аннотация: The problem of nonparametric identification of linear dynamic objects is investigated. In contrast with parametric identification, the case is analyzed when the equations describing the dynamic object is not specified up to parameters. Moreover, the identification problem is analyzed under normal object operation, opposite to the pПоказать полностьюreviously known nonparametric identification based on the Heaviside function input to the object and further Duhamel integral application. Voluntary signal is input to the object during normal operation and the weight function realizations are represented by observations of input and output object variables. Further it can be estimated as a nonparametric statistic such as Nadarya-Watson regression function. This nonparametric models are applicable to linear dynamic processes with different depths of memory. The identification problem when a deltashaped signal is input to linear object is also investigated. Nonparametric algorithms are given in both cases, when delta-shaped signal is input to the object and voluntary signal is input to the object under normal operation conditions. Statistical computer modeling methods confirm the efficiency of proposed algorithms. Some numerical studies are given in the article.
Журнал: Aspire to Science
Номера страниц: 166-170
Издатель: Новосибирский государственный технический университет