Steady-state data-driven dynamic stability assessment in the Korean power system.

Area of Science:

• Electrical Engineering
• Power Systems Analysis
• Computational Intelligence

Background:

• Dynamic security assessment (DSA) and stability prediction in power systems traditionally rely on high-resolution post-fault data, which are difficult and costly to acquire.
• The impracticality of widespread phasor measurement unit (PMU) deployment limits the use of high-resolution data in real-world scenarios.
• Existing methods often treat stability prediction as a black box, lacking physical interpretability.

Purpose of the Study:

• To develop a rotor angle stability prediction model utilizing easily obtainable steady-state power grid data.
• To address the practical limitations of high-resolution data acquisition in dynamic security assessment.
• To enhance the interpretability and efficiency of power system stability prediction.

Main Methods:

• A novel framework integrating physical insights from the extended equal-area criterion with machine learning techniques.
• Feature data extraction strategies to reduce input dimensionality for support vector machine (SVM) models.
• Partitioning time-series power flow data by month to account for system topology variations and using 5-min interval data for training.

Main Results:

• The proposed framework effectively predicts rotor angle stability using steady-state pre-contingency data.
• Demonstrated effectiveness in real-time response to critical line fault events.
• The method successfully identified unstable cases and trained an SVM with extracted features.

Conclusions:

• Steady-state data can be effectively utilized for rotor angle stability prediction, offering a practical alternative to high-resolution data.
• The developed framework provides a more interpretable and computationally efficient approach to dynamic security assessment.
• This research offers a viable solution for enhancing the reliability and security of power systems in real-time.