Subsynchronous oscillation detection via cycle-based feature extraction and weighted kernel extreme learning machine

• Published in: Frontiers in energy research Vol. 10
• Main Authors: Wang, Lei, Li, Tiecheng, Fan, Hui, Hu, Xuekai, Yang, Lin, Yang, Xiaomei
• Format: Journal Article
• Language: English
• Published: Frontiers Media S.A 20.09.2022
• Subjects: cycle; detection; feature extraction; subsynchronous oscillations; weighted kernel extreme learning machine
• ISSN: 2296-598X; 2296-598X
• DOI: 10.3389/fenrg.2022.964811

The fast and reliable detection of subsynchronous oscillations (SSOs) in wind farms is important for maintaining the stable operation of a power system. This paper presents a novel, SSO detection method for noisy synchrophasor data that considers the issue of detection as a binary classification (SSO and non-SSO) from the perspective of pattern recognition. The proposed algorithm easily implements cycle-based feature extraction from raw data by applying the strong period of the SSO signal, which is distinguished from noisy data. To mitigate the performance reduction of regular classifiers due to the imbalance issue caused by SSO data being substantially less than non-SSO data, a weighted kernel extreme learning machine is constructed as a classifier to implement SSO detection. Experimental studies are carried out on simulation and field data; the results show the effectiveness of the proposed algorithm for SSO detection in the case of a low SNR and of imbalance issues.