Технология настройки прогнозной модели на основе ретроспективного анализа собственных фазовых траекторий
Дата
2015
ORCID
DOI
Науковий ступінь
Рівень дисертації
Шифр та назва спеціальності
Рада захисту
Установа захисту
Науковий керівник
Члени комітету
Назва журналу
Номер ISSN
Назва тому
Видавець
НТУ "ХПИ"
Анотація
Предложен метод параметрического синтеза прогнозной модели на основе ретроспективного анализа собственных фазовых траекторий внутреннего параметра модели, который позволяет осуществлять обоснованный выбор значений настроечного параметра в соответствии с внутренней динамикой, выраженной в виде фазовых портретов, тем самым повышая точность прогнозирования по сравнению с традиционными поисковыми подходами к выбору параметра прогнозной модели. Предлагаемый подход к параметрическому синтезу прогнозной модели позволяет аналитически оценить качество модели в ходе ее использования, тем самым формализуя процесс параметрической настройки прогнозной модели.
It is considered the problem of the development of methods for parametric synthesis of predictive models, taking into account the peculiarities of the real state of the statistical information stored in the form of time series. The subject of research is the methods of parametric synthesis of one-parameter predictive models. The purpose is the expansion of methodological tools for parametric synthesis of the above models. It is proposed the method of parametric synthesis of predictive model based on a retrospective analysis of their own phase trajectories of internal model parameter, which allows for an informed choice of values of tuning parameters in accordance with the internal dynamics expressed in the form of phase portraits, thereby increasing prediction accuracy compared to traditional search approach to the selection of predictive model parameter. The proposed approach to the parametric synthesis of predictive model allows analytically assess the quality of the model in the course of its use, there by formalizing the process of parametric configuration of predictive model. It is constructed a diagram of decomposition process for parametric synthesis of predictive model. The proposed method can be used as a prediction value in the time series, and in the process of selecting appropriate predictive model to predict the specific time series.
It is considered the problem of the development of methods for parametric synthesis of predictive models, taking into account the peculiarities of the real state of the statistical information stored in the form of time series. The subject of research is the methods of parametric synthesis of one-parameter predictive models. The purpose is the expansion of methodological tools for parametric synthesis of the above models. It is proposed the method of parametric synthesis of predictive model based on a retrospective analysis of their own phase trajectories of internal model parameter, which allows for an informed choice of values of tuning parameters in accordance with the internal dynamics expressed in the form of phase portraits, thereby increasing prediction accuracy compared to traditional search approach to the selection of predictive model parameter. The proposed approach to the parametric synthesis of predictive model allows analytically assess the quality of the model in the course of its use, there by formalizing the process of parametric configuration of predictive model. It is constructed a diagram of decomposition process for parametric synthesis of predictive model. The proposed method can be used as a prediction value in the time series, and in the process of selecting appropriate predictive model to predict the specific time series.
Опис
Ключові слова
фазовый анализ, временные ряды, РФП, phase analysis, retrospective phase portrait
Бібліографічний опис
Романенков Ю. А. Технология настройки прогнозной модели на основе ретроспективного анализа собственных фазовых траекторий / Ю. А. Романенков // Вестник Нац. техн. ун-та "ХПИ" : сб. науч. тр. Темат. вып. : Механико-технологические системы и комплексы. – Харьков : НТУ "ХПИ". – 2015. – № 36 (1145). – С. 47-52.