Robust State Estimation Model for Low Voltage Distribution Networks in the Presence of Multiple Gross Errors

Abstract

One of the primary challenges in solving the State Estimation (SE) problem in low voltage networks is the presence of Gross Errors (GE). If the SE model fails to accurately estimate state variables in the presence of GE, the system operator may receive a distorted image of the power network, potentially leading to unforeseen disruptions, blackouts, and significant economic losses. The classical weighted least squares method, which is commonly used in such problems, exhibits low accuracy when simultaneous GE occurs. This paper provides a comprehensive analysis comparing robust M-estimators designed to handle GE, such as Hachtel and the largest normalized residual test, and presents a novel SE method called the Adaptive Maximum Correntropy Criterion (AMCC). The AMCC employs the maximum correntropy criterion for the loss function and an interior point methodology. Additionally, an adaptive scheme is employed to automatically adjust a high parameter associated with the shape of the related gamma function. We show that, in a real low voltage network, the AMCC exhibits superior accuracy with smaller root-mean-square errors compared to the other estimators studied.

Keywords

• Cauchy
• Hachtel
• interior point method
• largest normalized residual test
• least square
• least absolute value