Identification of power grids low-frequency oscillations through a combined MEEMD-Prony method

Abstract

Currently, the primary focus of power systems is to enhance precision and resilience against interference in identifying low-frequency oscillation modes. This research proposes a new method combined with ensemble empirical mode decomposition (MEEMD) to improve the sensitivity of traditional Prony to noise in parameter analysis of low-frequency oscillation signals. The method decomposes the measurement signal into intrinsic mode functions (IMF) using MEEMD, introduces permutation entropy to detect randomness, and reconstructs the remaining IMF components. The reconstructed signal was analyzed using the Prony method to extract the characteristic parameters associated with the low-frequency oscillation frequency. Simulations using numerical signals and the EPRI-36 node system confirm the method effectively suppresses modal mixing, mitigates noise interference, and accurately identifies low-frequency oscillation parameters. This approach offers advantages over traditional methods, including enhanced resistance to noise and increased accuracy.