UDC 004.827
Authors
A.O. NEDOSEKIN, Z.I. ABDULAEVA, S.V. ALEXANDROV
Abstract
Purpose of work. Demonstrate the capabilities of fuzzy models for interpreting highly noisy data obtained during testing of power plants. Method. For analysis, fuzzy-parabolic regression is used, which best approximates the statistics accumulated during the tests. The parameters of the regression curve are triangular fuzzy numbers that are identified by the maximum likelihood criterion, using the divergence functional F and keeping its value to a minimum. Result. Fuzzy-regression models are presented for two factors by which statistics were collected during the tests. In one of the cases, the model turned out to be piecewise nonlinear, built on two fuzzy polynomials of the first and second order. Conclusion The presence of a regression model of statistical tests allows the transition from identification models of a power plant to prognostic ones, making a forecast also in fuzzy form.
Keywords
diesel power plants, model of fuzzy parabolic regression, signal, noise, triangular fuzzy regression.
