Home / Current Issue / Paper 1717927
Energy Theft Detection in Smart Grids Using Machine Learning: A Structured Review and Proposed Hybrid Framework
Subject area: Science,Engineering and Technology · Area of research: ML for Smart Grid Energy Theft Detection
DOI: https://doi.org/10.64388/IREV9I11-1717927
Abstract
Electricity theft is a critical challenge for global power utilities, causing annual losses estimated at USD 96 billion through meter tampering, illegal connections, and cyber manipulation of smart meter data. Traditional rule-based and manual inspection methods are inadequate against increasingly sophisticated theft techniques in modern smart grid environments. This paper presents a structured review of 25 IEEE-indexed publications (2022–2025) on machine learning-based energy theft detection, analysing methodologies, datasets, model architectures, and key limitations. Thematic classification identifies six major research directions: supervised classification, deep learning, hybrid and ensemble methods, IoT-integrated systems, anomaly detection, and satellite and geospatial AI approaches. Systematic gap analysis reveals ten critical deficiencies — including absence of real-time deployment, scalability constraints, limited temporal modelling, lack of automated alerts, and dataset imbalance — and maps fifteen targeted research questions and objectives. A novel intelligent hybrid ML-based energy theft detection framework is proposed, aimed at improving detection accuracy, enabling real-time automated alerting, and minimising electricity losses in smart grid environments.
Keywords
Energy Theft Detection, Non-Technical Losses, Smart Grid, Machine Learning, Deep Learning, LSTM, Anomaly Detection, SVM, Ensemble Learning, IoT, Smart Meter, SMOTE.
How to cite this paper
@article{1717927,
author = {Shaik Heera, Dr. Haripriya V},
title = {Energy Theft Detection in Smart Grids Using Machine Learning: A Structured Review and Proposed Hybrid Framework},
journal = {Iconic Research And Engineering Journals},
year = {2026},
volume = {9},
number = {11},
pages = {2551-2557},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1717927.pdf},
abstract = {Electricity theft is a critical challenge for global power utilities, causing annual losses estimated at USD 96 billion through meter tampering, illegal connections, and cyber manipulation of smart meter data. Traditional rule-based and manual inspection methods are inadequate against increasingly sophisticated theft techniques in modern smart grid environments. This paper presents a structured review of 25 IEEE-indexed publications (2022–2025) on machine learning-based energy theft detection, analysing methodologies, datasets, model architectures, and key limitations. Thematic classification identifies six major research directions: supervised classification, deep learning, hybrid and ensemble methods, IoT-integrated systems, anomaly detection, and satellite and geospatial AI approaches. Systematic gap analysis reveals ten critical deficiencies — including absence of real-time deployment, scalability constraints, limited temporal modelling, lack of automated alerts, and dataset imbalance — and maps fifteen targeted research questions and objectives. A novel intelligent hybrid ML-based energy theft detection framework is proposed, aimed at improving detection accuracy, enabling real-time automated alerting, and minimising electricity losses in smart grid environments.},
keywords = {Energy Theft Detection, Non-Technical Losses, Smart Grid, Machine Learning, Deep Learning, LSTM, Anomaly Detection, SVM, Ensemble Learning, IoT, Smart Meter, SMOTE.},
month = {May},
doi = {https://doi.org/10.64388/IREV9I11-1717927}
}