Real-Time ZIP Load Parameter Tracking Using Sensitivity-Based Adaptive Window and Variable Elimination With Realistic Synchrophasor Data

Situational awareness and decision support for the power grid require accurately representing all the grid components in the modeling and analysis tools. Accurate load modeling is critical to understand the impact of voltage-dependent load behavior. The traditional constant power load model is incapable of accurately considering the load behavior. Real-time data-driven power system applications can be easily realized with the availability of synchronized data at a high reporting rate from phasor measurement units (PMUs). PMU data at substation buses can be used for the accurate estimation of aggregated load model parameters. In this work, two novel algorithms, adaptive least squares assisted by variable elimination (

A-LSVE

) and recursive least squares with reconfiguration of covariance matrix (

RLS-RC

), are proposed for load parameter estimation.

A-LSVE

uses least-squares assisted by sensitivity-based variable elimination and adaptive selection of estimation windows. It uses the current window and two past neighboring data windows with a similar load to voltage sensitivity for parameter estimation.

RLS-RC

, uses recursive least squares with the sensitivity-based reconfiguration of the estimated covariance matrix. A preprocessing framework is developed for measurement data, which uses an ensemble-based bad data detector to mitigate the impact of data anomalies and a cascade filter (median filter followed by the Kalman filter) to reduce the effect of noise in PMU data in parameter estimation. Moreover, a confidence factor computation framework is proposed to give insight regarding the accuracy of estimated parameters. The developed algorithms are validated using realistic load parameter estimation data with noise and anomalies for the IEEE test system modeled using electromechanical and electromagnetic simulators.