Methods for optimizing data processing based on fuzzy adjustment of time series elements and identification model variables

The problem of optimal identification and processing of random time series (SVR) based on the properties of statistical, dynamic, and fuzzy models is formulated. A method for qualitative identification of SVR is proposed, which includes algorithms for fuzzy equations, logical inferences, taking into account the effects of environmental factors and non-stationary processes. A generalized algorithm for identifying SVR with adjustment and correction of variable values based on the rules of fuzzy logic, methods for searching for extrema by t -norms and s -norms is developed. Tools are designed for optimal data processing by determining an adequate model; parametric and structural identification of objects; search optimization; model training; identification of the “input and output” relationship; formation and use of a knowledge base, as well as sets of fuzzy rules, linguistic variables, membership functions, and algorithms for regulating variable values. Methods of fuzzy correction of distorted information by controlling the error of SVR identification are developed, and a software package is implemented that provides high accuracy of data processing with significantly lower costs.

Ҳали таржима қилинмаган