editor@bjddjournal.org +91 8600071634, +91 98903 82132
Open-Access, Transparent Peer-Reviewed, Refereed

BODHIVRUKSHA JOURNAL OF DIVERSE DISCIPLINE

ISSN: 3139-1486 (Online) | ISSN: 3139-9819 (Print) Impact Factor : 6.21 ESTD Year : 2025
Bi-monthly Multilingual Academic Research Multidisciplinary

*** Important Announcement: Call for Papers! Submissions are now open for the upcoming issue. Please visit the "Submit Paper" section to learn more. ***
← Back to Special Issue on Emerging Frontiers in Artificial Intelligence and Multidisciplinary Intelligent Systems

IoT and Machine Learning for Real-Time Energy Theft Detection in Smart Grids : A Review of IoT-Based Monitoring and Machine Learning Approaches

Department of Artificial Intelligence and Data Science Ajeenkya DY Patil School of Engineering, Pune , India · India

Abstract

Electricity theft continues to create practical and financial difficulties for power utilities. It contributes to non-technical losses, reduces revenue, makes demand estimates less dependable, and can affect the operation of distribution networks. Inspections and simple threshold-based checks remain useful in some situations, but they can be slow when a utility must monitor a large number of consumers. This review examines how the Internet of Things (IoT), smart meters, edge computing, and machine learning (ML) could be combined to screen electricity-use data and identify patterns that deserve closer attention. Smart meters record consumption over time, and the readings can be sent to an edge gateway or a utility platform. Before analysis, the data need to be checked and prepared so that a model can work with meaningful features. The approaches considered include support vector machines, random forests, neural networks, CNN/LSTM models, hybrid methods, active learning, and techniques for dealing with imbalanced datasets. The paper also describes a conceptual system design and explains how a model score and a selected threshold can be used to raise an alert. Precision, recall, F1-score, false-positive rate, and detection latency are discussed as measures of performance. Local processing may help reduce response time and the amount of detailed data sent to a central system; central platforms, meanwhile, can support wider analysis and model maintenance. In practice, the results will depend on data quality, dependable labels, secure communication, and responsible handling of customer information. An alert should prompt a review, not be treated as proof of theft. For this reason, IoT–ML detection is considered here as a support tool that needs careful testing and human oversight. Keywords: Internet of Things, smart grid, machine learning, electricity theft detection, smart meters, edge computing, anomaly detection, data privacy.

Keywords

Internet of Thingssmart gridmachine learningelectricity theft detectionsmart metersedge computinganomaly detectiondata privacy.

How to Cite This Article

Sarthak Deore,Dr. Divya Sharma. (2026). IoT and Machine Learning for Real-Time Energy Theft Detection in Smart Grids : A Review of IoT-Based Monitoring and Machine Learning Approaches. Special Issue on Emerging Frontiers in Artificial Intelligence and Multidisciplinary Intelligent Systems.

Rights

Copyright
BODHIVRUKSHA JOURNAL OF DIVERSE DISCIPLINE