An Information Theoretic Approach to Power-Substation Level Dynamic State Estimation With Non-Gaussian Noise

This letter presents an information-theoretic approach to implement a decentralized substation based dynamic state estimator. Field phasor measurement unit (PMU) data is generally corrupted with colored and non-Gaussian noise, owing to which the conventional techniques based on second-order statistics provide strongly biased state estimates. The breakdown point of the proposed correntropy based estimator is theoretically infinite, making it an ideal choice for filtering non-Gaussian errors and outliers of high magnitude.