Applicability and Limitation Analysis of PMU Data and Phasor Concept for Low- and High- Frequency Oscillations

Phasor Measurement Units (PMUs) convert high-speed waveform data into low-speed phasor data, which are fundamental to wide-area monitoring and control in power systems, with oscillation detection and localization among their most prominent applications. However, representing electrical waveform signals with oscillations using PMU phasors is effective only for low-frequency oscillations. This paper investigates the root causes of this limitation, focusing on errors introduced by Discrete Fourier Transform (DFT)-based signal processing, in addition to the attenuation effects of anti-aliasing filters, and the impact of low reporting rates. To better represent and estimate waveform signals with oscillations, we propose a more general signal model and a multi-step estimation method that leverages one-cycle DFT, the Matrix Pencil Method, and the Least Squares Method. Numerical experiments demonstrate the superior performance of the proposed signal model and estimation method. Furthermore, this paper reveals that the phasor concept, let alone PMU phasors, can become invalid for waveform signals with high-frequency oscillations characterized by asymmetric sub- and super-synchronous components. These findings highlight the fundamental limitations of PMU data and phasor concept, and emphasize the need to rely on waveform data for analyzing high-frequency oscillations in modern power systems.

Adversarial Purification for Data-Driven Power System Event Classifiers with Diffusion Models

The global deployment of the phasor measurement units (PMUs) enables real-time monitoring of the power system, which has stimulated considerable research into machine learning-based models for event detection and classification. However, recent studies reveal that machine learning-based methods are vulnerable to adversarial attacks, which can fool the event classifiers by adding small perturbations to the raw PMU data. To mitigate the threats posed by adversarial attacks, research on defense strategies is urgently needed. This paper proposes an effective adversarial purification method based on the diffusion model to counter adversarial attacks on the machine learning-based power system event classifier. The proposed method includes two steps: injecting noise into the PMU data; and utilizing a pre-trained neural network to eliminate the added noise while simultaneously removing perturbations introduced by the adversarial attacks. The proposed adversarial purification method significantly increases the accuracy of the event classifier under adversarial attacks while satisfying the requirements of real-time operations. In addition, the theoretical analysis reveals that the proposed diffusion model-based adversarial purification method decreases the distance between the original and compromised PMU data, which reduces the impacts of adversarial attacks. The empirical results on a large-scale real-world PMU dataset validate the effectiveness and computational efficiency of the proposed adversarial purification method.