Low-Complexity Dimensionality Reduction for Big Data Analytics in the Smart Grid
M. Mohajeri, A. Ghassemi, Thomas Aaron Gulliver · 2020
A polar projection-based algorithm is proposed to reduce the computational complexity of dimensionality reduction in unsupervised learning algorithms. In particular, we consider the K-means clustering algorithm. A new distance metric is developed to account for peak power consumption to cluster consumer load profiles. This is used to cluster load profiles according to both total and peak power consumption. Numerical results are presented which demonstrate a significant reduction in computational complexity compared to K-means clustering using conventional dimension reduction techniques.