ADAPTIVE CONTROL METHOD AND CRITERIA FOR VERIFYING THE RELIABILITY OF COMPUTATIONAL IMPLEMENTATION OF MODELS OF DYNAMIC SYSTEMS

Abstract

The issues of studying dynamic systems are most acute in the development, operation and modernization of control systems for various objects and processes. A characteristic feature of modern control systems is the widespread use of various types of computer systems that implement methods and algorithms for modeling and control. Therefore, the quality of functioning of modeling and control systems is largely determined by the characteristics of the computing resources used.The problem of ensuring the quality of functioning, in particular, of control systems with a computer in the control loop is very complex from both a technical and mathematical point of view. In solving this problem, practice has put forward one of the first places a scientific direction related to the development of methods and tools that ensure the reliability and validity of computing and information processing processes. In a number of applied problems, in particular when assessing the reliability of the obtained solution, the dominant value is the forecast time (i.e. the speed of its receipt) and the simplicity of the computational procedures involved. In particular, such requirements may arise when solving control problems of fast-moving dynamic processes (objects), especially in conditions of multivariate formulation (such problems may include: optimization problems with iterative procedures for finding the optimal control law; dispatch control problems in abnormal and emergency situations, when the best solution according to a certain criterion is sought from possible alternative ones; problems that are solved in an “accelerated” time scale, when the laws of program control are sought, provided that there is no possibility to organize closed-loop control systems, etc.). It is obvious that to check the reliability of calculations regarding the dynamic state of the studied objects and processes, it is rational to use control algorithms. Moreover, in the case of limited time for calculations, it is advisable, as a control (verification) algorithm, to use a simplified algorithm for solving the problem of assessing the reliability of the obtained solution to the initial modeling or control problem.