The development of probabilistic and statistical methods for identification of noisy transient processes of synchronous machines

Noisy transient processes (TPs) in the stator windings of synchronous machines (SMs) in experiments on sudden symmetrical short circuits (SCs), field dissipation, voltage recovery, shock excitation, and other issues are especially vulnerable when they are identified by the results of bench tests using the operating standards for testing of synchronous machines. The processing methods of such transient processes according to domestic and foreign standards are overloaded with labor-intensive graphic procedures and calculations using oscillogram data. These methods to date have not made it possible to achieve the desired accuracy of TP processing due to a considerable scatter of the results of their identification. The experiment concerning a sudden SC is the main test for all transient processes, as it is asymmetric and contains the largest number of current components in the SM stator windings. Consequently, this experiment is potentially promising for finding ways to provide a high accuracy and reliability of TP identification. The other above-listed TPs are symmetrical without an asymmetric component, and so their exact identification is provided in a way similar to in the sudden SC experiment. The developed probabilistic and statistical methods (PSMs) of TP identification in many respects solve existing problems. The article presents new possibilities for the development of these methods with effective use of variational series of a random sign with the detected nucleus of effective point samples. These capabilities increase the accuracy and reliability of identification results of these TPs and reduce the labor-intensiveness of studies of a random sign in the investigated range of a TP with a transition component when processing long-term TPs of powerful SMs. © 2017, Allerton Press, Inc.