Open AccessElectricity stealing time recognition method based on difference and K-means clustering
Author(s) -
Yining Yang,
Xue Yang,
Cai Hui,
Pan Boliang,
Jianyu Chen,
Jing Li,
Qian Guo
Publication year - 2021
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/647/1/012075
Subject(s) - cluster analysis , electricity , computer science , value (mathematics) , power (physics) , diagram , artificial intelligence , engineering , electrical engineering , machine learning , database , physics , quantum mechanics
The time of stealing electricity is an important element of stealing electricity. In order to solve the disadvantages of traditional manual determination of stealing time and make the verified stealing users pay electricity charges accurately, this paper proposes a new kind of method to determine the time of stealing electricity. In this method, the difference and clustering algorithms are combined through the electricity fluctuation rate. When the power fluctuation is small, the stealing time is determined through detecting an abnormal value with the difference distribution combined box diagram method; when the power fluctuation is large, the low power period of several consecutive days is found by clustering to determine the stealing time. The example demonstrates the high accuracy of this method and its value in practical applications.