OPTIMIZATION OF METHODS OF VERIFICATION FOR UNSTEADY TRAFFIC MODELS

Abstract

The purpose of the work is to identify the main features of the application of a typical algorithm for solving verification problems, built on an algorithm and a general verification method using a logical-mathematical model of unsteady traffic, constructed using the quantization method and Markov approximation. The multidimensional law of distribution of all values of the aggregated traffic in the nonstationarity interval is assumed Gaussian. To assess the accuracy of the approximation of traffic, the concept of reference traffic is introduced and a comparison of the first two moment characteristics of the bypass model and reference traffic is used. This allows you to compare multi-dimensional Gaussian distributions of reference and model unsteady traffic. The canonical decomposition of a bypass according to the degree of a polynomial was chosen as a reference model of the bypass aggregated traffic. To verify the obtained models, it is proposed to use target functionals, which determine the accuracy and reliability of traffic modeling, as criteria for the adequacy of the reference traffic model. Studies have shown that quantization and Markov approximation lead to the expectation model of unsteady traffic as a linear combination of exponentials. In order to ensure the adequacy of model traffic of reference traffic, quantized values should be considered as controlled variables. At the same time, the initial probabilities of the states and intensity of traffic change act as uncontrollable variables - constants solving optimal tasks that must be determined by the initial data. Attention should be paid to non-linear effects associated with the use of expansion in a row in exponential functions and the use of traffic change intensities as exponent indicators. The application of the proposed system of adequacy criteria allows for a comprehensive check of the adequacy of the models, it also makes it possible to carry out an assessment of the influence of uncontrollable variables on optimal solutions. Verification by the model verification method can significantly improve the reliability of systems for critical applications — many errors turn out to be in the early stages of the production cycle, which improves the quality of development.