Browse by author
Lookup NU author(s): Professor Vladimir TerzijaORCiD
Full text for this publication is not currently held within this repository. Alternative links are provided below where available.
© 1969-2012 IEEE.Recently, wideband oscillation events have frequently occurred in renewable power systems and made the system unstable. As a result, a method to identify the occurrence of the unstable oscillation event is needed. The existing methods usually use a long window to ensure high identification accuracy, while a long identification time will be introduced. To deal with this problem, this paper proposes a fast and accurate identification method based on a data-driven method, i.e., the cubic spline interpolated fast Fourier transform-random forest (CSIPFFT-RF) method. The CSIPFFT is used to preprocess the waveform data and extract the amplitude variation features of the oscillation component quickly, and the feature data is used to train the RF to identify and classify the oscillations. In simulation tests, the proposed method is compared with the methods that use the traditional FFT and zero-padding FFT to preprocess the data, and the methods that use the back propagation, long short-term memory, and extreme learning machine to identify the oscillation. Results show that the proposed method has much higher accuracy (over 99%) and shorter identification time (about 1 s) than the existing methods under various scenarios. Importantly, the proposed method still has high performance even in untrained scenarios, especially when real data with dynamic harmonics, noise, and missing values are used.
Author(s): Gao L, Chen L, Xie X, Terzija V, Chen Z
Publication type: Article
Publication status: Published
Journal: IEEE Transactions on Power Systems
Year: 2025
Pages: epub ahead of print
Online publication date: 26/11/2025
Acceptance date: 02/04/2018
ISSN (print): 0885-8950
ISSN (electronic): 1558-0679
Publisher: Institute of Electrical and Electronics Engineers Inc.
URL: https://doi.org/10.1109/TPWRS.2025.3637236
DOI: 10.1109/TPWRS.2025.3637236
Altmetrics provided by Altmetric
