Scientific Journal of KubSAU

The task of nonparametric identification of sequence objects with discrete-continuous nature of the process under nonparametric uncertainty, i.e. in conditions where a priori information is not sufficient for an informed choice of a model structure up to parameters is considered. Among series-connected objects, there can be objects both dynamic and instantaneous ones with a lag. This kind of technological chains is common in various industries, particularly in metal, power, oil refining, etc. in solving this problem were used methods of nonparametric identification theory, mathematical statistics and statistical modeling. The theory of nonparametric systems is based on local approximation methods, in particular algorithms for nonparametric estimation of different kind of dependency from observation of input-output variables of the object. The article presents a nonparametric model for the group of spinning objects with delay. In the work we show in detail the results of numerical studies showing that the use of nonparametric algorithms allows predicting process performance with sufficient accuracy