Fast Shapelet Learning for Power System Dominant Instability Mode Identification
- Resource Type
- Conference
- Authors
- Zhang, Runfeng; Shi, Zhongtuo; Yao, Wei; Huang, Yanhao; Tang, Yong; Wen, Jinyu
- Source
- 2021 International Conference on Power System Technology (POWERCON) Power System Technology (POWERCON), 2021 International Conference on. :2033-2037 Dec, 2021
- Subject
- Engineering Profession
Power, Energy and Industry Applications
Training
Learning systems
Time series analysis
Rotors
Power system stability
Feature extraction
Data models
simulation data analysis
data mining
transient instability
short-term voltage instability
shapelet learning
genetic algorithm
- Language
- ISSN
- 2642-6226
Digital simulation is an essential part of supporting the safe and stable operation of power systems. One of the most crucial processes is to make the control decision table after a large amount of transient stability simulation. This process requires to be able to accurately distinguish between the transient (rotor angle) instability and short-term voltage instability. This paper proposes a fast shapelet learning method to extract features from the original voltage and rotor angle curves and then classify them through a machine learning (ML) model. Shapelet learning is a powerful data mining method, which can be used to extract hidden explainable information from power system time-domain simulation data. To overcome the huge time complexity of the shapelet transformation method, a genetic algorithm is applied to speed up the solution procedure. A case study is conducted on an 8-machine 36-bus system, and the simulation result indicates that the proposed method is with high explainability and fast learning speed.