Electricity Theft Detection Using Deep Neural Network

Abstract

Electricity theft is a critical issue faced by power utilities, resulting in significant non-technical losses, revenue reduction, and decreased efficiency of power distribution systems. This paper proposes a deep learning-based approach for detecting electricity theft using a Deep Neural Network (DNN). The system utilizes real-time data collected from smart meters, which is preprocessed and analyzed to extract meaningful consumption patterns. These patterns are fed into a trained DNN model to identify abnormal usage behavior that may indicate theft. The proposed system enables automatic detection and alert generation, reducing the need for manual inspection and improving detection accuracy. Compared to traditional methods, it offers faster response, higher reliability, and better scalability. The results demonstrate that the system effectively detects fraudulent activities, ensuring fair billing and supporting the development of a secure and efficient smart grid.

Introduction

This paper presents an Electricity Theft Detection System based on Deep Neural Networks (DNNs) to address the growing problem of electricity theft, which causes significant financial losses and inefficiencies in power distribution systems. Traditional detection methods, such as manual inspections and rule-based approaches, are often slow, inaccurate, and difficult to scale. With the adoption of smart grids and Advanced Metering Infrastructure (AMI), large volumes of real-time consumption data can now be analyzed using artificial intelligence techniques to improve theft detection.

The proposed system collects electricity consumption data from smart meters, preprocesses it through cleaning, normalization, and feature extraction, and then analyzes it using a trained DNN model. The model learns patterns of normal and abnormal electricity usage and identifies suspicious consumption behaviors that may indicate theft. When anomalies are detected, the system automatically generates alerts for administrators and displays them on a monitoring dashboard.

The system consists of several integrated modules, including an Admin Module for monitoring and managing theft cases, a User Module for viewing consumption and billing information, an Alert Module for real-time notifications, a Bill Payment Module for secure online payments, and a Database Management Module using MongoDB to store user data, consumption records, billing details, and alerts. All components are connected through a centralized backend that enables real-time data processing and communication.

Experimental evaluation using real-world smart meter data demonstrated strong performance. After preprocessing and training on historical consumption records, the DNN model achieved an overall accuracy of 95%, along with high precision, recall, and F1-score, effectively distinguishing between normal and fraudulent electricity usage while minimizing false detections. The results indicate that the proposed approach is reliable, scalable, and suitable for real-time deployment in smart grid environments.

The study concludes that integrating deep learning with smart meter infrastructure significantly improves electricity theft detection, reduces manual effort, and enhances power system security. Future improvements include expanding datasets, adopting advanced models such as LSTM and CNN-LSTM, integrating IoT-enabled smart meters, deploying the system on cloud platforms, and incorporating blockchain technology to enhance security, transparency, and data integrity.

Conclusion

Electricity theft is a serious issue faced by power distribution companies worldwide, leading to significant non-technical losses, financial instability, and reduced reliability of electricity supply.Traditional methods such as manual inspection and rule-based detection are time-consuming, less accurate, and highly dependent on human intervention. To overcome these limitations, this project proposed an intelligent electricity theft detection system using a Deep Neural Network (DNN). The developed system collects electricity consumption data from smart meters and processes it through data preprocessing techniques such as noise removal, normalization, and feature extraction. Important consumption patterns, including daily usage trends, load variations, and abnormal spikes, are analyzed to identify suspicious behavior. The Deep Neural Network model is trained using labeled historical data to learn the difference between normal and fraudulent consumption patterns. After training, the model is tested on unseen data to evaluate its performance using metrics such as accuracy, precision, recall, and F1-score. The proposed approach improves detection accuracy and significantly reduces the need for manual inspection. It enables real-time monitoring and automated alert generation, allowing utility authorities to take immediate action against suspected theft cases. Additionally, the system promotes fair billing practices and enhances transparency in power distribution. Overall, the implementation of a DNN-based electricity theft detection system contributes to a smarter, more secure, and efficient energy management infrastructure. In the future, this system can be further enhanced by integrating IoT-enabled smart grids, advanced cybersecurity mechanisms, and large-scale deployment in real-world utility networks.

References

[1] M. Anwar, N. Javaid, A. Khalid, M. Imran, and M. Shoaib, “Electricitytheft detection using pipeline in machine learning,” in Proc. Int. Wireless Commun. Mobile Comput. (IWCMC), Jun. 2020, pp. 2138–2142. [2] Z. Zheng, Y. Yang, X. Niu, H.-N. Dai, and Y. Zhou, “Wide and deep convolutional neural networks for electricity-theft detection to secure smart grids,” IEEE Trans. Ind. Informat., vol. 14, no. 4, pp. 1606–1615, Apr. 2018. [3] Q. Louw and P. Bokoro, “An alternative technique for the detection and mitigation of electricity theft in South Africa,” SAIEE Afr. Res. J., vol. 110, no. 4, pp. 209–216, Dec. 2019. [4] Y. Peng, Y. Yang, Y. Xu, Y. Xue, R. Song, J. Kang, and H. Zhao, “Electricity theft detection in AMI based on clustering and local outlier factor,”IEEE Access, vol. 9, pp. 107250–107259, 2021. [5] A. Aldegheishem, M. Anwar, N. Javaid, N. Alrajeh, M. Shafiq, and H. Ahmed, “Towardssustainable energy efficiency with intelligent electricity theft detection in smart grids emphasising enhanced neural networks,” IEEE Access, vol. 9, pp. 25036–25061, 2021. [6] S. Notley and M. Magdon-Ismail, “Examining the use of neural networks for feature extraction: A comparative analysis using deep learning, support vector machines, and K-nearest neighbor classifiers,” 2018,arXiv:1805.02294. [7] M. Billah and S. Waheed, “Minimum redundancy maximum relevance(mRMR) based feature selection from endoscopic images for automatic gastrointestinal polyp detection,” Multimedia Tools Appl., vol. 79,nos. 33–34, pp. 23633–23643, Sep. 2020. [8] C. Moler. (Mar. 3, 2023). Splines and Pchips. [Online].G. Dong and H. Liu, Feature Engineering for Machine Learning and Data Analytics. Boca Raton, FL, USA: CRC Press, 2018. [9] B. D. Fulcher and N. S. Jones, “Highly comparative feature-based time series classification,” IEEE Trans. Knowl. Data Eng., vol. 26, no. 12,pp. 3026–3037, Apr. 2014. 44ELECTRICITY THEFT DETECTION USING DNN [10] M. Di Martino, F. Decia, J. Molinelli, and A. Fern´andez, “Improving electric fraud detection using class imbalance strategies,” in Proc. ICPRAM,2012, pp. 135–141 [11] B. Vega-M´arquez, I. Nepomuceno-Chamorro, N. Jurado-Campos, and C.Rubio-Escudero, “Deep learning techniques to improve the performance of olive oil classification,” Frontiers Chem., vol. 7, p. 929,Jan. 2020. [12] K. T. Chui, D. C. L. Fung, M. D. Lytras, and T. M. Lam, “Predicting at-risk university students in a virtual learning environment via a machine learning algorithm,” Comput. Hum. Behav., vol. 107, Jun. 2020, Art. no. 105584. [13] W. Jia, C. Xiu-Yun, Z. Hao, X. Li-Dong, L. Hang, and D. Si-Hao, “Hyper parameter optimization for machine learning models based on Bayesian optimization,” J. Electron. Sci. Technol., vol. 17, no. 1, pp. 26–40, 2019. [14] D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,”2017, arXiv:1412.6980. [15] M. K. Uc¸ar, M. Nour, H. Sindi, and K. Polat, “The effect of trainingand testing process on machine learning in biomedical datasets,” Math.Problems Eng., vol. 2020, pp. 1–17, May 2020. [16] M. N. Hasan, R. N. Toma, A.-A. Nahid, M. M. M. Islam, and J.-M. Kim,“Electricity theft detection in smart grid systems: A CNN-LSTM based approach,” Energies, vol. 12, no. 17, p. 3310, 2019. [17] P. Dangeti, Statistics for Machine Learning. Birmingham, U.K.: Packt Publishing, 2017 [18] H. Abdi and L. J. Williams, “Principal component analysis,” Wiley Inter discipl. Rev., Comput. Statist., vol. 2, no. 4, pp. 433–459, 2010. [19] S. Ngamchuen and C. Pirak, “Smart anti-tampering algorithm design for single phase smart meter applied to AMI systems,” in Proc. 10th Int. Conf.Electr. Eng./Electron., Comput., Telecommun. Inf. Technol., May 2013, pp. 1–6. [20] B. Khoo and Y. Cheng, “Using RFID for anti-theft in a Chinese electrical supply company: A cost-benefit analysis,” in Proc. Wireless Telecommun.Symp. (WTS), Apr. 2011, pp. 1–6.

Copyright

Copyright © 2026 Anjana M A, Aiswarya P R, Nandana Thilakan, Ruvayya P H, Ms. Sreema E R. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.