Identification of Low-Frequency Oscillation Modes Using PMU Based Data-Driven Dynamic Mode Decomposition Algorithm

Mohd Zuhaib,Mohd Rihan

doi:10.1109/access.2021.3068227

Mohd Zuhaib, Mohd Rihan

Open Access

https://doi.org/10.1109/access.2021.3068227

Copy DOI

Abstract

Power system inter-area oscillations curtail the power transferring capabilities of the transmission lines in a large interconnected power system. Accurate identification of dominant modes and associated contributing generators is important to avoid power system failures by taking appropriate remedial measures. This paper proposes a multi-channel Improved Dynamic Mode Decomposition (IDMD) algorithm-based modal analysis technique using Synchrophasors measurement. First, a reduced-order dynamic power system model is estimated and using this model dominant oscillation modes, corresponding modes shapes, damping ratio, coherent group of generators, participation factors are determined. To improve the accuracy data stacking technique is used to capture detailed information of the system. An optimal hard threshold technique is utilized to select the most optimal model order to avoid uncertainties due to the presence of high level of measurement noise. The study results show that the proposed algorithm gives an accurate and robust solution even in systems having high level of noise in the measurement data. The performance of the proposed technique is tested on simulated data from two-area four-machine system and wNAPS 41-bus 16-generator system with PMU measurements corrupted with different levels of measurement noise. To further strengthen the viewpoint, the proposed method is validated on real-time PMU measurement from ISO New England data to validate the accuracy of the proposed work.

Highlights

Low-frequency electromechanical oscillations exist inherently in any large interconnected power system
The damping of inter-area oscillation is most important as it affects the stability of large interconnected power system and limits the power-transferring capacity of the transmission lines exchanging power between two areas [3], [4]
The introduction of data stacking technique and optimal hard threshold technique for determination of reduced model order substantially improves the performance of standard Dynamic mode decomposition (DMD) algorithm in identifying the system dominant modes even in the presence of high level of measurement noise

Summary

INTRODUCTION

Low-frequency electromechanical oscillations exist inherently in any large interconnected power system. [22] utilizes an optimized DMD algorithm to extract low-frequency oscillation modes from wide-area measurement datasets using variable projection and finite difference style approximation method. Utilization of data stacking technique to increases the dimensions of the measurement matrix This increases the number of estimated eigenvalues that help in identifying the detailed dynamics of the low-frequency oscillation modes for better characterization of system dynamics. To address this issue data stacking technique using the hankelization concept is adopted to create time-shifted copies of datasets This increases the row dimensions of the measurement matrix to obtain the large eigenvalues of the system to capture the detailed system dynamics [28], [29]. The value of s varies from 200 to 450

OPTIMAL TRUNCATION OF SINGULAR VALUES USING HARD THRESHOLD TECHNIQUE

FREQUENCY AND DAMPING RATIO

PARTICIPATION FACTOR

TEST CASES

Findings

CONCLUSION