Answers are grounded in this paper and cite their sources.
Wide-area oscillation events can cause severe disruption to energy production and delivery in power systems. The use of data-driven approaches for oscillation detection and source location is gaining significant attention in power system studies. However, such oscillation events are rare, and detailed system recordings are not publicly accessible. They are mostly anonymously shared with the community, which makes it challenging to study and analyze the different scenarios of forced oscillation events. The lack of detailed system recordings from oscillation events has been a serious shortcoming in developing data-driven methods for oscillation analysis. This work has created an extensive set of time-stamped forced oscillation datasets called “Dataset for Power SYstem Forced Oscillation Responses (DPSYFOR)” (pronounceddecipher) using realistic power grid models to leverage data availability with a wide range of forced oscillation frequencies and sources. A reduced-order model as well as a large-scale validated model, depicting the system conditions of the August 10, 1996, blackout of the Western American power grid, have been used to generate the oscillation datasets. This resource will significantly support the development of data-driven algorithms, particularly machine learning approaches, which have demonstrated considerable potential in the power industry for event detection and classification. Researchers can also utilize these datasets to validate and tune the performance of their existing data-driven methods.
No takes yet. Share an insight, caveat, or question.
Joshi et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: