Recommended Citation
August 1, 2013.
Abstract
Given time series data from multiple power generators containing measurements of phase angle taken at many points in time. The goal is to cluster together generators which have similar phase angle behavior.
In order to achieve this goal, Fast Fourier Transforms and Euclidean distances were used to quantify similarities between generators. A graph was created in which two generators were connected if they were sufficiently similar (known as a nearest neighbor graph). Using different nearest neighbor values, matrices were generated based on similarities between generators. Specific time intervals were taken from the large data set to assess the time dependency of the methods.
Different procedures were applied to the original generated matrices to produce even better clustering of the given objects. These procedures included Markov Clustering Algorithm, normalizing and squaring, and Markov chains. This project has implications toward the future power grid and model reduction of large networks.
Disciplines
Applied Mathematics | Mathematics
Mentor
Emilie Hogan
Lab site
Pacific Northwest National Laboratory (PNNL)
Funding Acknowledgement
This material is based upon work supported by the S.D. Bechtel, Jr. Foundation and by the National Science Foundation under Grant No. 0952013. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the S.D. Bechtel, Jr. Foundation or the National Science Foundation. This project has also been made possible with support of the National Marine Sanctuary Foundation. The STAR program is administered by the Cal Poly Center for Excellence in Science and Mathematics Education (CESaME) on behalf of the California State University (CSU).
Included in
URL: https://digitalcommons.calpoly.edu/star/216