Comparative study on estimation and identification of electromechanical oscillation parameters

What is it about?

This study compares the performance of the bandpass filtering method and the Prony algorithm in identifying parameters of power system electromechanical oscillation. The bandpass filtering method suppresses noise by selectively passing signals in a certain frequency range, while the Prony algorithm directly estimates the signal parameters. In simulations, we find that the Prony algorithm outperforms the bandpass filtering method in extracting signal features. To further improve the performance of the Prony algorithm, we introduced a genetic algorithm for global optimization. This study provides an important reference for the selection of parameter identification methods for power system oscillation analysis, which is of practical importance for improving system stability and security. In addition, this study provides a new perspective on the application of genetic algorithms in signal processing and a theoretical basis for further research in related fields.

Featured Image

Photo by Waranont (Joe) on Unsplash

Photo by Waranont (Joe) on Unsplash

Why is it important?

In modern interconnected power grids, the increasing scale and complexity of the operating environment pose great challenges to dynamic analysis. Electromechanical oscillations [1] have become a common problem in global power grids, leading to a serious threat to the safe and stable operation of power systems [2]. Therefore, ensuring the stability and security of power systems has become a key research area worldwide. Online parameter estimation based on the Kalman filter [3] and identification based on the Prony method [4] can enable real-time monitoring of power system states, thereby enhancing the stability of power systems. The Prony method was first proposed by Prony in 1795 for fitting parameters of exponential damped sinusoidal. Subsequently, Trudnowski et al. first used this method to identify the mode parameters of electromechanical oscillation in power systems[5]. Based on this approach, estimation of a set of modes using multiple signals [6] and Regularized Robust Recursive Least Squares (R3LS) [7] are proposed to solve the multi-signal mode estimation problem based on phasor measurement unit (PMU) in power systems. However, the Prony algorithm, commonly used for parameter identification, is sensitive to noise and may lead to errors in the extracted parameters [8]. To improve the quality of the oscillating signals, it is recommended to use a bandpass filter [9] that selectively transmits oscillating signals in a specific frequency range, suppressing unwanted frequencies. This paper presents a comparative study of two methods for identifying known power system signal parameters: the bandpass filtering method and the Prony algorithm [10]. This paper aims to optimize the parameters of electromechanical oscillation modes in power systems, which can enhance the power EEICE-2024 Journal of Physics: Conference Series 2849 (2024) 012003 IOP Publishing doi:10.1088/1742-6596/2849/1/012003 2 system’s ability to withstand external disturbances and reduce the risk of system oscillations and instability. The study provides a comprehensive understanding of the characteristics of the various parameter identification methods [11]. The bandpass filtering method suppresses noise, while the Prony algorithm directly estimates signal parameters. The comparative study can identify the limitations of different methods and suggest ways to improve them. For example, one could introduce a genetic algorithm into the Prony algorithm to optimize its parameters. The simulation results indicate that the Prony algorithm, with optimized parameters using a genetic algorithm, outperforms other methods in extracting signal features. This study provides insights into selecting appropriate parameter identification methods for power system oscillation [12] analysis, contributing to improving system stability and security.

Perspectives