Prediction of Signals in Control Systems Based on Fuzzy Time Series

Abstract

Currently, approaches to the analysis and forecasting of the state of complex technical objects based on time series using fuzzy and neural network methods and models are being actively developed. The study, based on the idea of using fuzzy time series, is to solve the problem of forecasting processes that reflect the state of complex systems. The results of the study make it possible to implement in technical control systems the procedure for predicting signals with various dynamic properties using technology based on fuzzy logic. At the same time, the choice of the optimal parameters of the prediction algorithm, depending on the dynamic characteristics (frequency band) of the predicted signal, can significantly improve the quality of control in various technical systems. The paper describes the solution to the forecasting problem, illustrating the possibility of taking into account the direct and indirect mutual influence of interconnected signal components based on the concept of fuzzy time series. Experimental studies have been carried out and the results of using the proposed method have been obtained.