Detecting electricity consumption data of power users based on a data platform

Read full text Summarise with VIRAT
Authors: Yuanshen Hong, Liang Yu, Jingjing Lu, Qianao Yu, Yiwei Yin, Hongchen Ying
Cited by: 0
References: 7

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Abstract Due to the lack of consideration for the similarity of data features on similar days, the application effect of electricity user consumption data is poor. To address this, a research method for detecting electricity user consumption data based on a data platform is proposed. Lagrange interpolation is used to fill in missing values in the electricity consumption data sequence. The K-means clustering algorithm is employed to cluster electricity user consumption data on similar days, grouping days with similar consumption characteristics into one category, thereby enabling in-depth analysis of consumption patterns under different categories. A data platform is used to first cluster similar days and then integrate multi-source electricity consumption data to solve the problem of duplicate conflicts caused by differences in data formats from different databases. A recurrent neural network is pre-trained using unlabeled data, and its parameters are retained. These parameters are then transferred to the target domain, and the model is fine-tuned using a small amount of real labeled data to achieve anomaly detection. Experimental results clearly show that the average anomaly detection precision rate is 88% after applying this method.

Fields of Study & Subjects

Energy Load and Power Forecasting Smart Grid Energy Management Electricity Theft Detection Techniques

Publication details

Journal / source
Journal of Physics Conference Series
Volume
3192
Issue
1
Pages
012024–012024
Publication date
1 March 2026
Language
EN