Title :
Power load classification based on spectral clustering of dual-scale
Author :
Mu Fu-lin ; Li Hong-yang
Author_Institution :
Customer Service Center, State Grid Chongqing Electr. Power Co., Chongqing, China
Abstract :
In the light of the one-sidedness of commonly used algorithms of power load classification caused by single similarity function, and the defects of these algorithm which have special requirements to the data space distribution and are easy to fall into local optimal solution, proposes a new electric power load classification algorithm. The algorithm first proposed a dual-scale similarity function base on the combination of Euclidean distance and the shape of the curve, thus to describe the similarity between the power load curves more accurately. Then cluster load curves according to the principle of spectral clustering, thus to make the algorithm not sensitive to the data distribution and data dimension, and to ensure the convergence to the global optimal solution. This algorithm can make more performance on classification of different power users, and has great significance to the implementation of the power user load control.
Keywords :
load regulation; Euclidean distance; data space distribution; dual-scale similarity function; electric power load classification algorithm; power load curves; single similarity function; spectral clustering; Algorithm design and analysis; Classification algorithms; Clustering algorithms; Euclidean distance; Load flow control; Power systems; Shape; Euclidean distance; power load classification; shape of the curve; similarity function; spectral clustering;
Conference_Titel :
Control Science and Systems Engineering (CCSSE), 2014 IEEE International Conference on
Print_ISBN :
978-1-4799-6396-6
DOI :
10.1109/CCSSE.2014.7224529