ADAPTIVE IDENTIFICATION OF NON-STATIONARY AIRCRAFT, ASSESSMENT OF THEIR STABILITY

Abstract

For the most part, various processes are multi-link inertial systems that have a certain non-stationarity to one degree or another. The presence of inertia leads to the fact that it is customary to consider processes as dynamic objects. Such dynamic non-stationary objects can be described by models, both in continuous time (for example, differential equations) and in discrete time (for example, difference equations). Ultimately, the choice of a discrete or continuous form of representation of a non-stationary object is determined by which mathematical apparatus is considered more appropriate by the researcher.
 In some cases, it is possible to consider the process as a static object, which allows to significantly simplify the analysis of such a non-stationary object. However, replacing a dynamic object with a static model is not always possible and leads to an uncontrolled error at the output.
 Non-stationarity is an important characteristic of the process as an object of management and is caused primarily by changes in process properties, under the influence of changes in environmental conditions, aging and wear of equipment, etc. Non-stationarity of the process significantly complicates the analysis and synthesis of control systems of this kind of objects, since the theory for control systems stationary in time is not applicable for non- stationary systems.
 Depending on the set goal and management tasks, process models can have different forms of presentation.
 The article examines various forms of presentation and research of non-stationary dynamic objects and their parameters from the standpoint of synthesis of the circuit of assessment and management of various processes, in particular the flight of an aircraft.
 The results of the analysis of the accuracy of identification of non-stationary objects, the adaptation of systems when combining or distributing in time the stages of "learning" and "management" in conditions of stochastic uncertainty are presented. The conditions for obtaining reliable estimates of the current values of object parameters during their measurement, processing and transmission in communication channels in real time are determined.