Adversarial Purification for Data-Driven Power System Event Classifiers with Diffusion Models
Authors:
Yuanbin Cheng,
Koji Yamashita,
Jim Follum,
Nanpeng Yu
Abstract:
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 perturbation…
▽ More
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.
△ Less
Submitted 13 November, 2023;
originally announced November 2023.
On Distribution Grid Optimal Power Flow Development and Integration
Authors:
Sarmad Hanif,
Rabayet Sadnan,
Tylor E. Slay,
Nawaf Nazir,
Shiva Poudel,
Bilal Bhatti,
Andy Reiman,
Jim Follum,
Joseph McKinsey,
Tarek Elgindy,
Rui Yang
Abstract:
Due to changes in electric distribution grid operation, new operation regimes have been recommended. Distribution grid optimal power flow (DOPF) has received tremendous attention in the research community, yet it has not been fully adopted across the utility industry. Our paper recognizes this problem and suggests a development and integration procedure for DOPF. We propose development of DOPF as…
▽ More
Due to changes in electric distribution grid operation, new operation regimes have been recommended. Distribution grid optimal power flow (DOPF) has received tremendous attention in the research community, yet it has not been fully adopted across the utility industry. Our paper recognizes this problem and suggests a development and integration procedure for DOPF. We propose development of DOPF as a three step procedure of 1) processing the grid, 2) obtaining a tractable solution, and 3) implementing multiple solution algorithms and benchmarking them to improve application reliability. For the integration of DOPF, we demonstrate how a DOPF federate may be developed that can be integrated in a co-simulation environment to mimic the real-world conditions and hence improve its practicality to be deployed in the field. To demonstrate the efficacy of the proposed methods, tests on IEEE 123-bus system are performed where the usage of tractable formulation in DOPF algorithm development and its comparison to the benchmark solution are demonstrated.
△ Less
Submitted 8 December, 2022;
originally announced December 2022.
