.png){kind=icon}
.png){kind=icon}
Return
Interpretable data-driven method for sub/super-synchronous oscillation stability evaluation
.png){kind=icon}
PRE
AI
DOI：10.1016/j.epsr.2024.111399.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The stability of sub/super-synchronous oscillation (SSO) in wind power integrated power systems is closely related to system operating conditions. To avoid SSO issues, it is necessary to conduct the online stability evaluation. The data-driven methods have the potential to achieve fast and accurate stability evaluations for large-scale power systems. However, the lack of interpretability poses a significant obstacle to its practical application. To address the issue, an interpretable data-driven method is proposed in this paper. Firstly, the SSO stability evaluation model is built based on neural network (NN) and the SSO stability boundary is derived. Then, the regularization and pruning techniques are adopted to streamline the model and activation functions are piecewise linearized to obtain SSO stability classification hyperplanes, so that the SSO stability rules can be extracted. Furthermore, the input sensitivity and SSO stability margin index are proposed to obtain more information on the current operating points. Finally, the proposed method is applied on a single-wind-farm system and a practical multiple-wind-farms system. The NN evaluation model is built and the SSO stability rules are extracted to clarify the evaluation logic. Compared with existing interpretable methods, the accuracy is greatly improved by the proposed method.
Keywords:
Interpretability
Data-driven method
Stability evaluation
Sub/super-synchronous oscillation
Journal
IF:
4.2
Papers:
1.1W
Citations:
2.2W