On detailed synchronous generator modeling for massively parallel dynamic state estimation

Synchronous generators are normally represented in a simplified fashion to reduce computational complexity in dynamic state estimation (DSE). In this paper a dynamic state estimator for a sixth-order synchronous generator model was developed on the massively parallel graphic processing units (GPU) to provide detailed and accurate Extended Kalman Filter (EKF) based estimation of the generator states. The estimation results are compared with the time domain simulation results on the CPU to demonstrate the accuracy of the proposed method. Also a speed-up of 10.02 for a 5120 generator system is reported.