BASIC PRINCIPLES AND ANALOGIES IN SOLVING THE PROBLEMS OF ADAPTATION, LEARNING, PATTERN RECOGNITION, OPTIMIZATION AND SIMILAR PROBLEMS OF FLIGHT CONTROL

Using the method of mathematical programming and stochastic approximation, the development of algorithms for solving the problems of adaptation, learning, pattern recognition, identification and other similar problems of aircraft control is shown. In control theory, the optimization task is reduced to determining the extrema of various quality functionals, and with the help of calculus of variations, this task is simplified to finding the extrema of some functions. In the general case, the quality functional is given by the mathematical expectation for some arguments of some functions of these arguments. The property of adaptation consists in continuously obtaining information and using it to improve working conditions or a quality criterion. In the general sense, adaptation is considered as successive approximation or iteration. When adding constraints to adaptation algorithms, the method of Lagrange multipliers is used. The basic principle of pattern recognition is to construct surfaces that divide hyperspace into a finite number of regions, each of which has its own image. And after the process of learning and "recognition" means that the known types will be correctly assigned to one of the known classes of objects. The article presents a pattern recognition algorithm. Identification is characterized as the determination of static and dynamic properties of the controlled process (object) and external disturbances acting on the process (object). The process (object) identification algorithm is presented. It is also about optimal filtering, which requires the use of an adaptive algorithm. The optimal weight function of the filter is presented. The article also considers the issue of dual control and presents a block diagram that implements this control.