Open AccessMODELS OF SYSTEMS OF QUASIPERIODIC PROCESSES BASED ON CYLINDRICAL AND CIRCULAR IMAGES
Author(s) -
V. R. Krasheninnikov,
Olga E. Malenova,
Olga E. Malenova,
A.Yu. Subbotin
Publication year - 2021
Publication title -
izvestiâ samarskogo naučnogo centra rossijskoj akademii nauk
Language(s) - English
Resource type - Journals
ISSN - 1990-5378
DOI - 10.37313/1990-5378-2021-23-1-103-110
Subject(s) - quasiperiodic function , object (grammar) , autoregressive model , process (computing) , autoregressive integrated moving average , spiral (railway) , series (stratigraphy) , noise (video) , computer science , autoregressive–moving average model , statistical physics , mathematics , time series , artificial intelligence , physics , mathematical analysis , statistics , image (mathematics) , machine learning , paleontology , biology , operating system
The behavior of objects in many practical situations has a quasiperiodic character - the presence of noticeable periodicity with random variations of quasiperiods. For example, noise and vibration of an aircraft engine, hydroelectric unit, seasonal and daily fluctuations in atmospheric temperature, etc. In this case, the object can have several parameters, therefore the object is described by a system of several time series, that is, several random processes. The emerging monitoring tasks (assessing the state of an object and its forecast) require setting a model of such a system of processes and identifying it for a particular object based on the results of its observations. In this paper, to represent a quasi-periodic process, an autoregressive model is used in the form of sweeps of several cylindrical or circular images along a spiral. Choosing the values of a small number of parameters of this model, one can describe and simulate a wide class of systems of quasiperiodic processes. The problem of identifying a model is considered, that is, determining the values of its parameters at which it, in a certain sense, best corresponds to the actually observed process. This problem is solved using a pseudo-gradient adaptive procedure, the advantage of which is its real-time operation with low computational costs.