H-models for non-inertia systems with a delay

Бесплатный доступ

The author considers a problem of the new class of processes with а «tubular» structure in the space of “input-output” variables. Modeling of these class processes differs highly from the generally accepted parametric models being a surface of the same space. A question of multidimensional systems modeling having small volume of learning samples is analyzed specially. To develop the trainable parametric models of «tubular» processes, the corresponding non- parametric indicator is introduced. A new class of trainable parametric models, and some results of the numerical investigation are presented.

Identification, non parametric algorithms, parametric model, a priori information, stochastic processes, h-модели, h-models

Короткий адрес: https://sciup.org/148176977

IDR: 148176977

Статья научная