A robust ALOHA and sequential clustering-based mode estimator for low frequency oscillations in power system using...

Area of Science:

• Power Systems Engineering
• Signal Processing
• Control Theory

Background:

• Accurate estimation of low frequency modes is vital for power system small signal stability.
• Existing methods like Total Least Square estimation of signal parameters via rotational invariance techniques (TLS-ESPRIT) require prior knowledge of mode numbers.
• Current model order estimation techniques are sensitive to noise and outliers in auto-correlation matrices, hindering real-time automation.

Purpose of the Study:

• To propose a robust mode estimation technique for precisely detecting low frequency modes in power systems, even with high variance noise and outliers.
• To overcome the limitations of existing model order estimation methods.
• To enhance the automation and reliability of power system stability analysis.

Main Methods:

• Annihilating filter-based low-rank Hankel matrix (ALOHA) technique to create a rank-deficient Hankel matrix, mitigating noise and outliers in Phasor Measurement Unit (PMU) signals.
• Sequential K-Mean++ clustering to segregate eigenvalues of the auto-correlation matrix into signal and noise subspaces, identifying prominent low frequency modes.
• Utilizing the estimated model order with TLS-ESPRIT for mode estimation.

Main Results:

• The proposed ALOHA technique effectively nullifies noise and outliers in PMU signals.
• Sequential K-Mean++ accurately detects the number of dominant low frequency modes by distinguishing signal and noise subspaces.
• The integrated approach provides robust low frequency mode estimation, validated through simulations and real-world data.

Conclusions:

• The developed technique offers a robust solution for low frequency mode estimation in power systems, outperforming existing methods in noisy and outlier-prone conditions.
• This approach enhances the accuracy and automation of small signal stability assessments.
• The method's effectiveness is confirmed across various synthetic and real-world power system scenarios.