State Estimator Based on the Maximum Correntropy Criterion for Islanded Microgrids
State Estimation, Islanded Microgrids, Maximum Correntropy Criterion, Bad Data Rejection, Phasor Measurements.
This dissertation presents a methodology for state estimation in islanded radial microgrids, incorporating steady-state frequency as a state variable. The focus is on dispatchable generators operating under droop control mode. The proposed estimator is framed as an optimization problem, utilizing the Maximum Correntropy Criterion (MCC) and considering nodal power injections as pseudo-measurements to ensure system observability. It integrates synchronized measurements from Phasor Measurement Units (PMUs) with generation data from dispatchable generators. Initially, an Optimal Power Flow (OPF) algorithm generates measurements and minimizes power losses. The performance of the proposed MCC-based approach is compared with a Weighted Least Squares-based estimator (WLS), demonstrating superior capability in rejecting bad data. This methodology is tested on a simulated 33-bus islanded microgrid, specifically focusing on scenarios with a single piece of bad data. When juxtaposed against the OPF results (assumed error-free), the MCC-based estimator exhibits a deviation of 2.43% compared to the 14.96% deviation noted with the WLS-based estimator.