Distilling provider-independent data for general detection of non-technical losses

Meira, Jorge Augusto; Glauner, Patrick; State, Radu; Valtchev, Petko; Dolberg, Lautaro; Bettinger, Franck; Duarte, Diogo

doi:10.1109/peci.2017.7935765

Cited by 34 publications

(27 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One possible explanation is that detection of frauds may have explanatory variables other than the energy consumption time series not considered in this work. As a matter of comparison, [9] achieves AUC of 0.729 using random forests and handcrafted feature engineering. Their dataset is similar to the one used in this work, although being bigger, filtered and preprocessed differently, without sliding evaluations performed.…”

Section: Settings and Resultsmentioning

confidence: 99%

“…Previous work on NTL detection has employed different approaches, types of inputs and dataset sizes. For instance, [9] focuses on feature engineering with random forest, logistic regression and support vector machine as classifiers. A survey on this field is presented in [5], citing also other approaches based on fuzzy systems, genetic algorithms, etc.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Recurrent Dynamical Projection for Time Series-Based Fraud Detection

Antonelo

State

2017

Artificial Neural Networks and Machine Learning – ICANN 2017

Self Cite

View full text Add to dashboard Cite

Abstract. A Reservoir Computing approach is used in this work for generating a rich nonlinear spatial feature from the dynamical projection of a limited-size input time series. The final state of the Recurrent neural network (RNN) forms the feature subsequently used as input to a regressor or classifier (such as Random Forest or Least Squares). This proposed method is used for fraud detection in the energy distribution domain, namely, detection of non-technical loss (NTL) using a real-world dataset containing only the monthly energy consumption time series of (more than 300K) users. The heterogeneity of user profiles is dealt with a clustering approach, where the cluster id is also input to the classifier. Experimental results shows that the proposed recurrent feature generator is able to extract relevant nonlinear transformations of the raw time series without a priori knowledge and perform as good as (and sometimes better than) baseline models with handcrafted features.

show abstract

Section: Settings and Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Recurrent Dynamical Projection for Time Series-Based Fraud Detection

Antonelo

State

2017

Artificial Neural Networks and Machine Learning – ICANN 2017

Self Cite

View full text Add to dashboard Cite

show abstract

“…Our proposed approach deals with the situation where there are only a small number of labeled samples. There are different fashion of detecting NTL among these work, for example, classification based [6,7,23,25], statistical analysis based [4,5], clustering based [24] or merely judgements with threshold values [3]. Most of these work took the advantage of the abundance of customer type labels.…”

Section: 3discussionmentioning

confidence: 99%

Non-Technical Loss Detection in Power Grids with Statistical Profile Images Based on Semi-Supervised Learning

Wang

2019

Sensors

View full text Add to dashboard Cite

In order to keep track of the operational state of power grid, the world's largest sensor systems, smart grid, was built by deploying hundreds of millions of smart meters. Such system makes it possible to discover and make quick response to any hidden threat to the entire power grid. Non-technical losses (NTLs) have always been a major concern for its consequent security risks as well as immeasurable revenue loss. However, various causes of NTL may have different characteristics reflected in the data. Accurately capturing these anomalies faced with such large scale of collected data records is rather tricky as a result. In this paper, we proposed a new methodology of detecting abnormal electricity consumptions. We did a transformation of the collected time-series data which turns it into an image representation that could well reflect users' relatively long term consumption behaviors. Inspired by the excellent neural network architecture used for objective detection in computer vision domain, we designed our deep learning model that takes the transformed images as input and yields joint featured inferred from the multiple aspects the input provides. Considering the limited labeled samples, especially the abnormal ones, we used our model in a semi-supervised fashion that is brought out in recent years. The model is tested on samples which are verified by on-field inspections and our method showed significant improvement.

show abstract

“…Feature selection is an important task for the identification of NTL. In Meira et al (2017), the features are divided into four categories with respect to time, geography, similarity of consumption profile and infrastructure. Random forest, logistic regression and SVM are tested with different proportion of NTL ranging from 10 to 90% across all four categories.…”

Section: Hybrid Techniquesmentioning

confidence: 99%

Performance analysis of machine learning classifiers for non-technical loss detection

Ghori

Imran

Nawaz

et al. 2020

J Ambient Intell Human Comput

View full text Add to dashboard Cite

Power companies are responsible for producing and transferring the required amount of electricity from grid stations to individual households. Many countries suffer huge losses in billions of dollars due to non-technical loss (NTL) in power supply companies. To deal with NTL, many machine learning classifiers have been employed in recent time. However, few has been studied about the performance evaluation metrics that are used in NTL detection to evaluate how good or bad the classifier is in predicting the non-technical loss. This paper first uses three classifiers: random forest, K-nearest neighbors and linear support vector machine to predict the occurrence of NTL in a real dataset of an electric supply company containing approximately 80,000 monthly consumption records. Then, it computes 14 performance evaluation metrics across the three classifiers and identify the key scientific relationships between them. These relationships provide insights into deciding which classifier can be more useful under given scenarios for NTL detection. This work can be proved to be a baseline not only for the NTL detection in power industry but also for the selection of appropriate performance evaluation metrics for NTL detection.

show abstract

Distilling provider-independent data for general detection of non-technical losses

Cited by 34 publications

References 15 publications

Recurrent Dynamical Projection for Time Series-Based Fraud Detection

Recurrent Dynamical Projection for Time Series-Based Fraud Detection

Non-Technical Loss Detection in Power Grids with Statistical Profile Images Based on Semi-Supervised Learning

Performance analysis of machine learning classifiers for non-technical loss detection

Contact Info

Product

Resources

About