Type of Culminating Activity
Dissertation - Boise State University Access Only
Doctor of Philosophy in Electrical and Computer Engineering
Electrical and Computer Engineering
John N. Chiasson, Ph.D.
Said Ahmed-Zaid, Ph.D.
Techniques for the parameter estimation of large synchronous machines are an important and developing field. These techniques, for example, lead to the accurate prediction of large synchronous generators response to transient instabilities. Consequently, regional electricity coordinating councils require power utility companies to conduct tests and furnish the parameters of these machines for stability studies. To facilitate the reliable testing of large synchronous machines this work presents two parameter estimation techniques. Firstly, an offline stand-still estimation technique is developed, where the synchronous machine is disconnected from the grid with the rotor locked at an arbitrary (but known) angle, and a test is conducted over a short period of time (several seconds). Secondly, an online parameter estimation technique is developed that uses the data recorded during regular steady-state operation of the synchronous machine, and during regular transient operation (shifting between operating points). Both parameter estimation techniques are based on nonlinear least-squares estimation and elimination theory. Therefore, the resulting algorithms are non-iterative techniques where the data is used to construct polynomials whose roots are related to the parameter values. These finite number of roots are all that is needed to determine the parameter set which minimizes a global error criterion.
Furthermore, the offline and the online parameter estimation techniques were tested in simulation and experimentally to demonstrate their efficacy. The offline technique was successfully shown to work experimentally furnishing the parameters of a salient pole synchronous machine, and predicting its response to a chirp signal. A comparable and standardized technique is the Stand-Still Frequency Response (SSFR), which could take more than 6 hours to conduct, versus several seconds for the offline technique developed here-in. On the other hand, the online technique was studied via simulations that model measurex ment errors and with an online experimental setup. The analysis showed that the online technique is sensitive to measurement errors. These results for the online technique are reported with recommendations for its future development.
Oteafy, Ahmed M. A., "Novel Parameter Identification Techniques for Large Synchronous Generators" (2011). Boise State University Theses and Dissertations. Paper 234.