Understanding Electricity-Theft Behavior via Multi-Source Data

Abstract

Electricity theft, the behavior that involves users conducting illegal operations on electrical meters to avoid individual electricity bills, is a common phenomenon in the developing countries. Considering its harmfulness to both power grids and the public, several mechanized methods have been developed to automatically recognize electricitytheft behaviors. However, these methods, which mainly assess users' electricity usage records, can be insufficient due to the diversity of theft tactics and the irregularity of user behaviors.In this paper, we propose to recognize electricity-theft behavior via multi-source data. In addition to users' electricity usage records, we analyze user behaviors by means of regional factors (non-technical loss) and climatic factors (temperature) in the corresponding transformer area. By conducting analytical experiments, we unearth several interesting patterns: for instance, electricity thieves are likely to consume much more electrical power than normal users, especially under extremely high or low temperatures. Motivated by these empirical observations, we further design a novel hierarchical framework for identifying electricity thieves. Experimental results based on a real-world dataset demonstrate that our proposed model can achieve the best performance in electricitytheft detection (e.g., at least +3.0% in terms of F0.5) compared with several baselines. Last but not least, our work has been applied by the State Grid of China and used to successfully catch electricity thieves in Hangzhou with a precision of 15% (an improvement from 0% attained by several other models the company employed) during monthly on-site investigation.

Fraud Detection on Power Grids While Transitioning to Smart Meters by Leveraging Multi-Resolution Consumption Data

“…Complex deep learning architectures show excellent results on this data, exceeding 90% precision [3], [6], [15]. This performance is overoptimistic when we compare it with the performance reported by distribution companies when novel inspections are performed based on the prediction of classification algorithms, with precisions ranging from 15% to 47% [16], [17]. This performance gap illustrates how important it is to have access to real data and proper segmentation (prior to the in-site inspections) to guide algorithm training.…”

“…In recent years deep learning architectures have been used showing promising results on smart meter data. A variety of deep learning solutions have been proposed based on convolutional neural networks (CNN), [3], [20], long short-term memory (LSTM) layers, and recurrent neural networks [16], [17], [21].…”

Treating Class Imbalance in Non-Technical Loss Detection: An Exploratory Analysis of a Real Dataset

“…A similar dataset from China is used in [20]. The authors have not dealt with the imbalance behavior of the data.…”

A framework to select a classification algorithm in electricity fraud detection

“…The last two decades have seen considerable research directed towards electricity fraud. From the earlier work by Galvan et al 10 who evaluated electricity usage in the Spanish farming sector, to Davidson 15 and Fourie and Calmeyer 16 who introduced NTL research into the South African context, to the behavioural identification of Hu et al 17 , the opportunity for research is extensive. Examples of research using fraud detection classification in electricity include Nizar et al 18 who used a NB classifier and decision tree algorithm to assess the consumption load profile of customers at different time intervals.…”