The Problem of 'Stored Energy' – the most Important Point in the Classification of Identification Errors

摘要

The problem of establishing dependencies based on empirical data with a random component has been known since the 18th century. In the works of Gauss and Laplace, a parametric class of functions and a functional are introduced, the minimization of which leads to the search for the required function. In the twentieth century, the development of the ideas of cybernetics and the appearance of computers made it possible to approach the solution of the problems of identifying the dynamic properties of objects from the data of their normal operation. However, here it was required to solve a range of systemically related problems. Their modern classification puts the problem of "stored energy" in the first place. The mechanism of occurrence of the corresponding error and the determining factors are considered. Analytical relationships for the resulting error are obtained. Verification was carried out using computer simulation.