Identification of Interarea Modes From Ambient Data of Phasor Measurement Units Using an Autoregressive Exogenous Model

Jin Kwon Hwang,Jeonghoon Shin

doi:10.1109/access.2021.3067213

Abstract

Various modal identification methods for interarea modes have been developed to improve identification accuracy by overcoming the measurement noise in ambient data of phasor measurement units (PMUs). In this study, a modal identification method that is insensitive to measurement noise is proposed by introducing bandpass filters, which extract a modal signal from an autocorrelation function of PMU ambient data. The bandwidth of the filters is set to be adequately narrow such that the noise can be rejected sufficiently. To reduce the computational burden of the proposed method, the filters are designed by transforming a reference lowpass filter. An autoregressive exogenous (ARX) model of interarea modes is applied to the extracted modal signal. The parameters of the ARX model are estimated for modal identification via the least squares method. The identification accuracy of the proposed method is compared with those of conventional modified extended Yule Walker and discrete Fourier transform methods with respect to the signal-to-noise ratio and modal damping by using synthetic ambient data. Finally, the feasibility of the proposed method is demonstrated by identifying interarea modes of Kundur's two-area four-machine system and two real power systems in South Korea.

Highlights

Interarea oscillations can degrade the transfer capacity and stability of a power system
We propose a modal identification method that is insensitive to measurement noise by using an autoregressive exogenous (ARX) model derived from an AF of ambient frequency data
In this study, a modal identification method insensitive to measurement noise has been proposed by using an ARX model, which is obtained through bandpass filtering of the AF of phasor measurement units (PMUs) ambient data

Summary

INTRODUCTION

Interarea oscillations can degrade the transfer capacity and stability of a power system. We propose a modal identification method that is insensitive to measurement noise by using an autoregressive exogenous (ARX) model derived from an AF of ambient frequency data. The proposed method utilizes oscillation powers around modal peaks of the DFT magnitude similar to the DFT method but does not cause spectral leakage of the DFT or mitigate the sensitivity to modal damping It uses the LSM in the time domain like the MEYW method but does not require spurious modes to cope with measurement noise. When there are K interarea modes in a power system, a modal signal y (t) in ambient frequency data is expressed as follows: K y (t) = k=1 qk (t) (1). The noise components in an AF of ambient frequency data should be rejected to enhance the identification accuracy of interarea modes To achieve this in the proposed method, an oscillation signal of the K modes is extracted from rw[m] through bandpass filtering. 0.0181 + 0.0543z−1 + 0.0543z−2 + 0.0181z−3 1 − 1.7600z−1 + 1.1829z−2 − 0.2781z−3

PERFORMANCE COMPARISON BY SIMULATIONS

MODAL IDENTIFICATION OF REAL POWER SYSTEMS

INLAND POWER SYSTEM

CONCLUSION