Least Squares Estimation-Based Synchronous Generator Parameter Estimation Using PMU Data

TL;DR

This paper introduces a least squares estimation method using ARX models to accurately identify key dynamic parameters of synchronous generators from PMU data, enhancing real-time grid stability analysis.

Contribution

It develops a novel ARX model-based LSE approach for generator parameter estimation, including detailed conversion methods and validation with numerical results.

Findings

Accurately estimates generator inertia and control droop from PMU data.

Demonstrates effectiveness of ZOH and Tustin conversion methods.

Provides a practical framework for real-time generator parameter identification.

Abstract

In this paper, least square estimation (LSE)-based dynamic generator model parameter identification is investigated. Electromechanical dynamics related parameters such as inertia constant and primary frequency control droop for a synchronous generator are estimated using Phasor Measurement Unit (PMU) data obtained at the generator terminal bus. The key idea of applying LSE for dynamic parameter estimation is to have a discrete \underline{a}uto\underline{r}egression with e\underline{x}ogenous input (ARX) model. With an ARX model, a linear estimation problem can be formulated and the parameters of the ARX model can be found. This paper gives the detailed derivation of converting a generator model with primary frequency control into an ARX model. The generator parameters will be recovered from the estimated ARX model parameters afterwards. Two types of conversion methods are presented: zero-order hold (ZOH) method and Tustin method. Numerical results are presented to illustrate the proposed LSE application in dynamic system parameter identification using PMU data.