Power big user segmentation method based on improved AP and K-means clustering

A k-means clustering and large-scale user technology, applied in data processing applications, character and pattern recognition, instruments, etc., can solve the problems of sensitive outliers, difficulty in determining the number of clusters k, easy to fall into local optimum, etc., to achieve profit high effect

CN107767293AInactive Publication Date: 2018-03-06ELECTRIC POWER RES INST OF STATE GRID ZHEJIANG ELECTRIC POWER COMAPNY +1
0 Cites 14 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Publication Date
2018-03-06
Estimated Expiration
Not applicable · inactive patent

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The invention discloses a power big user segmentation method based on improved AP and K-means clustering. The method is mainly targeted for power industry big users, and is used for carrying out segmentation on power big users through improved AP and K-means combined algorithms based on three index data of monthly electricity consumption, historical credit score and electricity consumption growthrate comprising ring growth rate and year-on-year growth rate of the big users. The power big user segmentation method carries out segmentation on the power big users based on combined clustering algorithms of an improved AP algorithm and an improved K-means algorithm, is good in segmentation effect, and prevents the defects that it is hard to determine cluster number k, it is sensitive to isolated points and it is easy to fall into local optimum when carrying out gradient descent search of a conventional K-means algorithm and the like.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The invention relates to a clustering algorithm for large electric power users based on the combination of improved AP and improved K-means, belonging to the field of power system analysis and calculation. technical background

[0002] With the continuous deepening of power market reform, the status of large power users in the power market has become increasingly prominent. By subdividing large users, power supply companies can further understand large users, determine large power user groups, and identify valuable large user behaviors and Value characteristics, so as to develop targeted service measures and differentiated marketing strategies, and significantly improve the service level of power supply enterprises while effectively controlling resource allocation costs [1]. The large power user groups proposed in the present invention mainly refer to industrial users in the electric power industry, which consume large amounts of electricity, have high...

Examples

Embodiment Construction

[0040] The implementation of the present invention will be further described below in conjunction with the accompanying drawings and examples, but the implementation and inclusion of the present invention are not limited thereto.

[0041] Because the data is large and difficult to handle, the present invention firstly performs normalization processing on all data before clustering, for the data set x 1 ,x 2 ,...,x n To transform:

[0042]

[0043] then y 1 ,y 2 ,...,y n ∈[0,1] and dimensionless.

[0044] First, use the AP clustering algorithm to process the data, input 3 or 4 indicators in the power consumption, credit score and power consumption growth rate of large users, and use the AP algorithm to calculate the initial clustering center and the initial clustering number K , use the initial cluster center and the number K of clusters calculated by the AP algorithm to find the initial value of the cluster for K-means clustering. Firstly, cluster analysis is carried...