Showing 1–2 of 2 results for author: Eftekharnejad, S

A Predictive Online Transient Stability Assessment with Hierarchical Generative Adversarial Networks

Authors: Rui Ma, Sara Eftekharnejad, Chen Zhong, Mustafa Cenk Gursoy

Abstract: Online transient stability assessment (TSA) is essential for secure and stable power system operations. The growing number of Phasor Measurement Units (PMUs) brings about massive sources of data that can enhance online TSA. However, conventional data-driven methods require large amounts of transient data to correctly assess the transient stability state of a system. In this paper, a new data-drive… ▽ More Online transient stability assessment (TSA) is essential for secure and stable power system operations. The growing number of Phasor Measurement Units (PMUs) brings about massive sources of data that can enhance online TSA. However, conventional data-driven methods require large amounts of transient data to correctly assess the transient stability state of a system. In this paper, a new data-driven TSA approach is developed for TSA with fewer data compared to the conventional methods. The data reduction is enabled by learning the dynamic behaviors of the historical transient data using generative and adversarial networks (GAN). This knowledge is used online to predict the voltage time series data after a transient event. A classifier embedded in the generative network deploys the predicted post-contingency data to determine the stability of the system following a fault. The developed GAN-based TSA approach preserves the spatial and temporal correlations that exist in multivariate PMU time series data. Hence, in comparison with the state-of-the-art TSA methods, it achieves a higher assessment accuracy using only one sample of the measured data and a shorter response time. Case studies conducted on the IEEE 118-bus system demonstrate the superior performance of the GAN-based method compared to the conventional data-driven techniques. △ Less

Submitted 14 December, 2021; originally announced December 2021.

A PMU-based Multivariate Model for Classifying Power System Events

Authors: Rui Ma, Sagnik Basumallik, Sara Eftekharnejad

Abstract: Real-time transient event identification is essential for power system situational awareness and protection. The increased penetration of Phasor Measurement Units (PMUs) enhance power system visualization and real time monitoring and control. However, a malicious false data injection attack on PMUs can provide wrong data that might prompt the operator to take incorrect actions which can eventually… ▽ More Real-time transient event identification is essential for power system situational awareness and protection. The increased penetration of Phasor Measurement Units (PMUs) enhance power system visualization and real time monitoring and control. However, a malicious false data injection attack on PMUs can provide wrong data that might prompt the operator to take incorrect actions which can eventually jeopardize system reliability. In this paper, a multivariate method based on text mining is applied to detect false data and identify transient events by analyzing the attributes of each individual PMU time series and their relationship. It is shown that the proposed approach is efficient in detecting false data and identifying each transient event regardless of the system topology and loading condition as well as the coverage rate and placement of PMUs. The proposed method is tested on IEEE 30-bus system and the classification results are provided. △ Less

Submitted 1 December, 2018; originally announced December 2018.