Improve of power system oscillation detection accuracy based on optimal sampling rate for data-driven real-time

Introduction. The rapid integration of renewable energy sources and inverter-based resources has increased the complexity of modern power systems, leading to higher risks of instability and low-frequency oscillations. Problem. While data-driven analysis using real-time measurements offers better adaptability than model-based methods, the use of high-resolution data leads to an exponential increase in analysis time, which hinders the effectiveness of real-time monitoring and rapid response. Goal. To improve power system oscillation detection accuracy. This goal is achieved on the basis of the developed method to determine the optimal sampling rate for real-time oscillation detection by performing a comparative performance evaluation of three major algorithms: Prony, MPM (Matrix Pencil Method), and HTLS (Hankel Total Least Squares). Methodology. To establish the optimal balance for highly accurate real-time applications, the study analyzed detection accuracy (frequency and damping ratio) and CPU execution time by varying sampling rates using simulated data from the IEEE 39-bus system and actual field measurement data from PMUs. Results. The analysis confirmed that highly accurate oscillation mode identification is successfully achieved at optimal sampling rates of 15 sps (samples per second) for real-world data and 5 sps for simulation data. The results demonstrated that while detection capabilities generally improve with higher rates, applying these specific optimal thresholds effectively suppresses noise and strictly balances the crucial trade-off between rigorous estimation accuracy and real-time processing speed. Scientific novelty. The research systematically demonstrates how optimizing the non-linear trade-off between sampling density and computational overhead directly enhances oscillation detection accuracy. By quantifying this relationship for Prony, MPM and HTLS algorithms, it establishes a novel boundary condition that guarantees highly accurate and reliable real-time applicability. Practical value. The providing technical guidelines for power system engineers to improve oscillation detection accuracy in wide-area monitoring systems. By applying the proposed optimal sampling rates, engineers can maximize detection fidelity and prevent false assessments without violating real-time constraints. References 21, tables 12, figures 8.