About the event
Student: Seyedeh Armina Foroutan
Degree: Ph.D. Electrical & Computer Engineering
Committee Chair: Dr. Anurag Srivastava
Dissertation Title: Generator Model Validation and Calibration using Synchrophasor Measurements
Abstract: Planning and operation of power grid highly depends on the analytical studies using models accurately representing the electric assets in the field. Model validation and calibration of generating units are important to ensure the compliance of the models with the actual behavior of the generating assets. Parameter calibration of these Grey-box models, is quite challenging and has not yet been solved in full generality. The traditional methods involve taking generators off-line for intrusive experimental validation and manual calibration. However, Phasor Measurement Units (PMUs) provide synchronized measurements with high sampling rates and create an opportunity to avoid offline testing and achieve a cost efficient solution. Online testing using synchrophasors usually involves one form of Bayesian based Filtering (BbF) approaches to estimate states and parameters simultaneously.
Traditionally, parameter calibration is associated with bulk generators, but with increasing penetration level of Distributed Energy Resources (DER), their dynamic behavior is no longer negligible. DER aggregators can be considered virtual power plants and are the candidates for model validation. Model parameter estimation is quite challenging due to the dependency between parameters and states causing numerical stability and convergence issues, large number of parameters, the quality of synchrophasors data and their integration with commercial dynamic modeling tools.
This work provides a detailed comparative analysis for conventional Kalman-based and Bayesian-based Filtering approaches with the associated numerical stability and convergence problem and proposed novel algorithms and architectures for online parameter calibration of generating units and integration process with commercial dynamic simulators. Results indicate superiority of the proposed dual extended Kalman Filter for generator model parameter estimation.