Comparing Electricity Consumer Categories Based on Load Pattern Clustering with Their Natural Types

• Published in: Algorithms and Architectures for Parallel Processing Vol. 10393; pp. 658 - 667
• Main Authors: Jiang, Zigui, Lin, Rongheng, Yang, Fangchun, Liu, Zhihan, Zhang, Qiqi
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2017; Springer International Publishing
• Series: Lecture Notes in Computer Science
• Subjects: Clustering; Consumer category; Consumer type; Load pattern; Smart Grid
• ISBN: 9783319654812; 3319654810
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-319-65482-9_51

As one aspect of smart city, smart gird has similar situation such as big data issue. Data analysis of daily load data generated by smart meters can benefit both electricity suppliers and end consumers. Electricity consumer categorization based on load pattern clustering is one of research subjects. This paper aims to achieve a better understanding of electricity consumer categorization by detecting the relationships among consumer categories and their natural types. A two-stage clustering based on multi-level 1D discrete wavelet transform and K-means algorithm is applied to perform daily load curve clustering and load pattern clustering. Additionally, to obtain distinct consumer categories, method of category identification based on association rule mining and characteristic similarity is also proposed in this paper. Experiment is conducted on data set of 24-value daily load data with labels of consumer types. Based on the comparison of experimental results, both relationships and differences exist among consumer categories and consumer types but consumer types cannot determine consumer categories.