Precise and reliable identification of the transient processes of a powerful synchronous machine by probabilistic statistical methods

Abstract

Identification of the transient process (TP) obtained by probabilistic and statistical methods based on probability theory and mathematical statistics as well as on the graphical–analytical method (GAM) in accordance with State Standard GOST 10169–77 was considered according to the results of bench tests of an STM-800-2.6 kV synchronous machine (SM) at an 4512-V test voltage during a sudden symmetrical shortcircuit (SSC). Despite the high labor intensity and low precision and reliability of TP oscillograms due to the numerous graphic operations and transitions from analogue TP to the discrete form and back, the GAM is a basic one and is widely used. Oscillogram processing by the GAM aiming at TP transition from analogue to discrete form leads to deviation of peaks from their original location. For this reason, practically no electromagnetic parameters adequately correspond to the TP of SM. The method of minimization of the meansquare error of approximation of the transient component to the experimental data was proposed. This error is presented in the form of the surface in a 3D coordinate system, for which reason the procedure of optimization, modeling, and determining the error in each node sampling with account taken of the influence of different random factors allow to reveal the minimum error. The method of determining the surge current by one vertex without defining TP parameters during SSC experience was tested as well. Error analysis of identification results of the TP SM by GAM and PSM for all phases is presented.