Open AccessOrthogonal method for solving maximum correntropy‐based power system state estimation
Author(s) -
Freitas Victor,
Costa Antonio Simões,
Miranda Vladimiro
Publication year - 2020
Publication title -
iet generation, transmission and distribution
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.92
H-Index - 110
eISSN - 1751-8695
pISSN - 1751-8687
DOI - 10.1049/iet-gtd.2019.1179
Subject(s) - outlier , robustness (evolution) , estimator , computer science , solver , algorithm , orthogonal array , electric power system , mathematical optimization , power (physics) , mathematics , artificial intelligence , taguchi methods , statistics , biochemistry , chemistry , physics , quantum mechanics , machine learning , gene
This study introduces a robust orthogonal implementation for a new class of information theory‐based state estimation algorithms that rely on the maximum correntropy criterion (MCC). They are attractive due to their capability to suppress bad data. In practice, applying the MCC concept amounts to solving a matrix equation similar to the weighted least‐squares normal equation, with difference that measurement weights change as a function of iteratively adjusted observation window widths. Since widely distinct measurement weights are a source of numerical ill‐conditioning, the proposed orthogonal implementation is beneficial to impart numerical robustness to the MCC solution. Furthermore, the row‐processing nature of the proposed solver greatly facilitates bad data removal as soon as outliers are identified by the MCC algorithm. Another desirable feature of the orthogonal MCC estimator is that it avoids the need of post‐processing stages for bad data treatment. The performance of the proposed scheme is assessed through tests conducted on the IEEE 14‐bus, 30‐bus, 57‐bus and 118‐bus test systems. Simulation results indicate that the MCC orthogonal implementation exhibits superior bad data suppression capability as compared with conventional methods. It is also advantageous in terms of computational effort, particularly as the number of simultaneous bad data grows.