Electricity Theft Detection in Smart Grids

Based on Deep Neural Network

SYNOPSIS

B. E. [Information Technology]

of

Savitribai Phule Pune University

By

Name of students

Under the guidance of

Name of guide
 Matoshri Education Society’s
MATOSHRI COLLEGE OF ENGINEERING AND RESEARCH CENTRE,
Near Odha Village, Nashik – Aurangabad Highway Nashik – 422 105, (M.S.), India

AcademicYear- 2023-24

Synopsis of Proposed Project for the degree of B.E. (Information

Technology)

Name of course
1 Project Stage-I ( 414448)

2 Project Group ID

Title of Project
3

Project Domain
4

Sponsorship Details( (Name, External

5 Guide name and Designation with
Signature, e- Mail ID), if any
Name of Guide
6

[Student’s name and sign] [Sign of guide]

Prof. N. L. Bhale
Head, Information Technology Department
 Abstract
The increasing demand for electricity has led to the growth of smart grids, which offer numerous
advantages such as improved energy efficiency, reduced power outages, and enhanced security.
However, one of the significant challenges in smart grids is electricity theft, which is a major cause of
revenue loss for utility companies. So, electricity theft is a major concern for electric power
distribution companies. The aim of this project is to develop an effective approach for detecting
electricity theft in smart grids based on Artificial Neural Network (ANN). The proposed approach will
use electricity usage dataset which is referred from the popular web repository kaggle. The collected
data will be preprocessed and fed into the ANN, which will learn to identify patterns and anomalies in
the consumption data. The ANN model will be trained using a dataset of legitimate consumption
patterns and then tested with data that contains instances of electricity theft. To evaluate the
performance of the proposed approach, the model will be tested on a test data. The results predicted
from our proposed system of electricity theft detection in smart grids using ANN is Good. Our system
achieved Training Accuracy of 99% and Validation Accuracy of 99%. The performance metrics used
will include accuracy, precision, recall, and F1-score. We also developed the proposed system in Flask
Web framework for easy usage with better User Interface for the predicting the results. The expected
outcome of this project is an effective approach for detecting electricity theft in smart grids using
ANN, which can be used by utility companies to improve their revenue collection and enhance the
security of the smart grid. This project can also be extended to other domains that involve anomaly
detection in large-scale datasets, such as fraud detection in financial systems and intrusion detection in
computer networks.
 SYNOPSIS OF THE WORK

1. Relevance and Introduction:

The proliferation of smart grids has revolutionized the way electricity is generated,
distributed, and consumed. This modernization has brought about numerous benefits,
including enhanced energy efficiency, real-time monitoring, and demand response
capabilities. However, along with these advancements, the challenge of electricity theft has
also evolved, necessitating innovative solutions to safeguard the integrity of the power
distribution infrastructure.
Electricity theft, often referred to as "energy pilferage," is a significant concern for utility
companies and regulators worldwide. It involves unauthorized consumption of electricity
without proper billing, resulting in substantial financial losses and compromised grid
reliability. Traditional methods of detecting theft, which primarily rely on manual
inspections and rule-based algorithms, have proven inadequate in addressing the growing
sophistication of fraudulent practices.
To tackle this issue, there is a pressing need for more advanced and automated techniques
that can accurately identify instances of electricity theft in real time. This project proposes
the utilization of deep neural networks (DNNs) as a cutting-edge approach to electricity theft
detection in smart grids. DNNs have demonstrated remarkable capabilities in various
domains, including image recognition, natural language processing, and anomaly detection.
The fundamental premise of this project revolves around harnessing the power of DNNs to
analyze the voluminous data generated by smart meters and other sensors within the grid. By
capturing intricate patterns, correlations, and anomalies hidden within this data, DNNs can
provide a more accurate and efficient means of distinguishing between legitimate energy
consumption and fraudulent activities. This approach holds the promise of significantly
improving the accuracy and effectiveness of theft detection systems.
In this context, the following sections of the project will delve into the methodology, data
collection and preprocessing, DNN architecture design, and experimental results. By
addressing the challenges associated with electricity theft using a state-of-the-art technology
like DNNs, this project aspires to contribute to the advancement of smart grid security and
reliability, ultimately benefiting both utility providers and consumers alike.
2. Literature review:
Research on electricity theft detection in smart grids has attracted many researchers to devise
methods that mitigate against electricity theft. Methods used in the literature can be broadly
categorized into the following three categories: hardware-based, combined hardware and
data-based detection methods and data-driven methods. Hardware-based methods [13]–[19]
generally require hardware devices such as specialized microcontrollers, sensors and circuits
to be installed on power distribution lines. These methods are generally designed to detect
electricity theft done by physically tampering with distribution components such as
distribution lines and electricity meters. They can not detect cyber attacks. Electricity cyber
attack is a form of electricity theft whereby energy consumption data is modified by hacking
the electricity meters [7]. For instance, in [13], an electricity meter was re-designed. It used
components that include: a Global System for Mobile Communications (GSM) module, a
microcontroller, and an Electrically Erasable Programmable Read-Only Memory
(EEPROM). A simulation was done and the meter was able to send a Short Message Service
(SMS) whenever an illegal load was connected by bypassing the meter. Limited to detecting
electricity theft done by physically tampering with distribution components such as
distribution lines and electricity meters. Authors in [16] used the GSM module, ARM-cortex
M3 processor and other hardware components to solve the electricity theft problem done in
the following four ways: bypassing the phase line, bypassing the meter, disconnecting the
neutral line, and tampering with the meter to make unauthorized modifications. A prototype
was built to test all four possibilities. The GSM module was able to notify with SMS for
each theft case. Authors in [17] designed ADE7953 chip-based smart meter which is
sensitive to current and voltage tempering, and mechanical tempering. ADE7953 was used
to detect overvoltage, dropping voltage, overcurrent, the absence of load and other
irregularities in voltage and current. It sent an interrupt signal to the Microcontroller Unit
(MCU) which reported tampering status. Mechanical tampering was overcome by
connecting a tampering switch to MCU’s IO ports so that it can send alarm signals to MCU
once tampered with. The design was tested with tampering cases such as connecting the
neutral and the phase lines, connecting the meter input and output in reverse mode, and
bypassing the phase line to load. The probability of detection failure was 2.13%. Authors in
[15] used a step down transformer, voltage divider circuit, microchip and other hardware
components to design a circuitry to detect electricity theft by comparing forward current on
the main phase line with reverse current on the neutral line. The circuitry was installed
before the meter.The design was tested on Proteus software and on actual hardware. When
the meter was bypassed, the problem was detected and an alarm sounded. In [14], a circuit to
detect electricity theft done by bypassing the meter was designed. The transformers,
rectifiers, microcontroller, GSM module and other hardware components were used. The
 GSM controller notified the operator with SMS when the meter was bypassed.
Authors in [25] employed a combination of CNN and Long Short-Term Memory (LSTM)
recurrent neural network deep learning techniques. Seven hidden layers were used, of which
four of them were used by CNN and three were utilized by LSTM. This method relied on
CNN’s automatic feature extraction ability on a given dataset. Features were extracted from
1-D time-series data. On model validation, the maximum accuracy achieved was 89%. The
authors in [26] used a combination of Local Outlier Factor (LOF) and k-means clustering to
detect electricity theft. They used k-means clustering to analyze the load profiles of
customers, and LOF to calculate the anomaly degrees of customers whose load profiles were
from their cluster centres. On the evaluation of the method, they attained an AUC of 81.5%.
Our model achieves a maximum value of 91.8% accuracy and 97% on validation. In [27],
two electricity theft models were developed. The first model is based on Light Gradient
Boosting (LGB) classifier. A combination of SMOTE and Edited Nearest Neighbour (ENN)
was used to balance the dataset. Feature extraction was done using AlexNet, followed by
classification with LGB. This proposed model was named as SMOTEENN-AlexNet-LGB
(SALM). The second model is based on the Adaptive Boosting classifier. Conditional
Wasserstein Generative Adversarial Network with gradient penalty (CWGAN-GP) was used
to generate synthetic data that resembled the minority class data to balance data of the
unbalanced classes. Feature extraction was performed using GoogleNet, then classification
by AdaBoost followed. The proposed model was named as GAN-NETBoost. The models
were evaluated with SGCC data used in this work. SALM and GAN-NetBoost attained an
accuracy of 90% and 95%, and AUC of 90.6% and 96% respectively on validation.
Although these models were able to achieve impressive results, their consideration of time-
domain features alone limited their performance. Our solution shows that adding frequency-
domain features on time-domain features improves classification performance.

3. Motivation
The motivation behind this project is to address the critical issue of electricity theft in smart
grids. As smart grids become more prevalent, traditional methods of theft detection are
becoming inadequate. This project aims to leverage the power of deep neural networks
(DNNs) to analyze the vast amount of data generated by smart meters and accurately
identify instances of theft. By doing so, it seeks to enhance grid reliability, reduce financial
losses for utility companies, and promote fair energy distribution.

4. Problem statement:
The project's core challenge is to develop an effective solution for detecting electricity theft
within smart grids. Conventional theft detection methods are insufficient due to the
complexity of modern grids. This project aims to employ deep neural networks (DNNs) to
 analyze smart meter data and identify abnormal consumption patterns indicative of theft,
ultimately enhancing grid security and financial integrity.

5. Objectives:
 To develop a customized deep neural network (DNN) model for precise electricity theft
detection in smart grids, leveraging advanced feature engineering and rigorous
evaluation metrics.
 To curate a comprehensive dataset encompassing genuine consumption data and
simulated theft scenarios, facilitating effective training and validation of the DNN
model.
 To optimize the DNN model's parameters, layer configurations, and hyperparameters
through systematic training, ensuring accurate identification of electricity theft
instances.
 To integrate the trained DNN model into a real-time monitoring system capable of
efficient analysis of incoming smart meter data while minimizing processing latency.
 To assess the DNN model's performance rigorously using metrics like accuracy,
precision, recall, and F1-score, showcasing its ability to discern theft instances and
minimize false alarms.
 To conduct a comparative analysis, highlighting the superiority of the DNN-based
approach over traditional methods in terms of electricity theft detection.
 To ensure the developed solution's scalability to handle large-scale smart grid
deployments and robustness against varying consumption patterns and environmental
conditions.
 To meticulously document the entire development process, from data preprocessing to
model architecture, and provide actionable insights for potential deployment.
 To emphasize the positive impact of the solution on utility companies, consumers, and
grid security by reducing revenue losses and promoting equitable energy distribution.

6. Scope of Project
The scope of this project encompasses the following aspects:
 Deep Neural Network Development: Design and develop a deep neural network (DNN)
model specifically tailored for electricity theft detection in smart grids.
 Data Collection and Preparation: Curate an extensive dataset comprising legitimate
consumption records and simulated theft scenarios, preparing the data for training and
evaluation.
 Feature Engineering: Implement advanced techniques for feature extraction to capture
 intricate consumption patterns and anomalies indicative of electricity theft.
 Model Training and Optimization: Train and fine-tune the DNN model using the curated
dataset, optimizing parameters, layers, and hyperparameters.
 Real-time Monitoring System: Integrate the trained DNN model into a real-time
monitoring system capable of analyzing smart meter data streams efficiently.
 Performance Evaluation: Assess the model's performance using metrics like accuracy,
precision, recall, and F1-score, validating its ability to detect theft while minimizing
false alarms.
 Comparative Analysis: Compare the DNN-based approach's performance with
traditional methods for electricity theft detection.
 Scalability and Robustness: Ensure the solution's scalability to accommodate large-scale
smart grid deployments and robustness against varying consumption patterns.
 Documentation and Reporting: Thoroughly document the project's lifecycle, from data
preprocessing to model architecture, and provide insights for potential implementation.
 Impact and Benefits: Emphasize the positive impact on grid security, utility companies,
and consumers by reducing revenue losses and promoting equitable energy distribution.

7. List of required hardware, software, or other equipment for

executing the project:
Hardware Requirement
• Processor (Intel Dual Core) : 2 GHz
• RAM : 4 GB
• Hard Disk : 256 GB (Min)
• IO Devices : Mouse, Keyboard

Software Requirement
• Operating System : Windows 7 Onwards
• Coding language : Python (Version 3.8)
• Web Framework : Flask
• IDE : VS Code

8. Proposed system and expected outcomes:

The proposed system seeks to revolutionize electricity theft detection within smart grids by
harnessing the power of advanced deep neural networks (DNNs). This comprehensive
solution encompasses data processing, model development, training, real-time monitoring,
and performance evaluation. Initially, data collected from smart meters, including historical
consumption patterns and real-time readings, undergoes preprocessing to ensure
compatibility with the DNN model. The heart of the system lies in the DNN architecture,
meticulously designed to handle complex consumption patterns and identify anomalies
indicative of theft. Through iterative training and optimization, the model's parameters and
hyper-parameters are fine-tuned to achieve heightened accuracy in detecting instances of
electricity theft while minimizing false alarms. Integrated into a real-time monitoring
system, the trained DNN continuously analyzes incoming smart meter data streams,
promptly alerting utility companies upon detecting potential theft instances. The system's
performance is rigorously evaluated using established metrics, including accuracy, precision,
recall, and F1-score, providing quantifiable evidence of its effectiveness. Additionally, a
comparative analysis with traditional methods underlines the superiority of the DNN-based
approach. Ultimately, the expected outcomes encompass improved accuracy, reduced false
alarms, real-time monitoring capabilities, scalability, and a significant technological
advancement in the domain of critical infrastructure security. By contributing to enhanced
grid security, financial integrity, and equitable energy distribution, the proposed system
holds the potential to transform the landscape of smart grid operations.

Expected Outcomes:
 Enhanced Theft Detection Accuracy: The DNN-based approach is anticipated to yield
higher accuracy in detecting electricity theft instances compared to conventional
methods, thereby reducing revenue losses for utility companies.
 Reduced False Alarms: Through rigorous training and optimization, the DNN model
should minimize false alarms, ensuring that legitimate consumption patterns are not
incorrectly flagged as theft.
 Real-time Monitoring Capability: The integrated real-time monitoring system should
enable prompt identification of theft instances, allowing utility companies to take swift
action.
 Scalability and Robustness: The solution's scalability and robustness will ensure its
effectiveness in handling varying consumption patterns and environmental conditions in
large-scale smart grid deployments.
 Quantifiable Metrics: By using standard evaluation metrics, the project will provide
quantifiable evidence of the DNN model's performance in detecting theft, allowing for
objective assessment.
 Technology Advancement: The successful implementation of the proposed system will
 contribute to the advancement of using deep neural networks in critical infrastructure
security, specifically within the context of smart grid operations.
 Operational Efficiency: The solution's ability to accurately detect theft and unauthorized
consumption will lead to improved operational efficiency for utility companies, as well
as a fairer energy distribution system.

9. Architecture Diagram:

10. Future scope:

In comparison with other data-driven methods evaluated on the same dataset, we obtained
97% AUC which is 1% higher than the best AUC in existing works, and 91.8% accuracy,
which is the second-best on the benchmark. The method used here utilizes consumption
data patterns. Apart from its application in power distribution networks, it can be used in
anomaly detection applications in any field. Our work brings a small contribution towards
accurately detecting energy theft as we detect theft that only took place over time. We wish
to extend our method to detect real-time electricity theft in the future. Since this method
was evaluated based on consumption patterns of SGCC customers, it can further be
validated against datasets from different areas to ensure its applicability anywhere.

11. References:
[1] S. Foster. (Nov. 2, 2021). Non-Technical Losses: A $96 Billion Global
Opportunity for Electrical Utilities. [Online]. Available:
https://energycentral.com/c/pip/ non-technical-losses-96-billion-globalopportunity-
electrical-utilities
[2] 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.
[3] M. Anwar, N. Javaid, A. Khalid, M. Imran, and M. Shoaib, ‘‘Electricity theft
detection using pipeline in machine learning,’’ in Proc. Int. Wireless Commun.
Mobile Comput. (IWCMC), Jun. 2020, pp. 2138–2142.
 [4] 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.
[5] P. Pickering. (Nov. 1, 2021). E-Meters Offer Multiple Ways to Combat Electricity
Theft and Tampering. [Online]. Available:
https://www.electronicdesign.com/technologies/meters
[6] X. Fang, S. Misra, G. Xue, and D. Yang, ‘‘Smart grid—The new and improved
power grid: A survey,’’ IEEE Commun. Surveys Tuts., vol. 14, no. 4, pp. 944–980,
4th Quart., 2012. [7] M. Ismail, M. Shahin, M. F. Shaaban, E. Serpedin, and K.
Qaraqe, ‘‘Efficient detection of electricity theft cyber attacks in AMI networks,’’ in
Proc. IEEE Wireless Commun. Netw. Conf. (WCNC), Apr. 2018, pp. 1–6.
[8] A. Maamar and K. Benahmed, ‘‘Machine learning techniques for energy theft
detection in AMI,’’ in Proc. Int. Conf. Softw. Eng. Inf. Manage. (ICSIM), 2018, pp.
57–62.
[9] A. Jindal, A. Schaeffer-Filho, A. K. Marnerides, P. Smith, A. Mauthe, and L.
Granville, ‘‘Tackling energy theft in smart grids through data-driven analysis,’’ in
Proc. Int. Conf. Comput., Netw. Commun. (ICNC), Feb. 2020, pp. 410–414.
[10] I. Diahovchenko, M. Kolcun, Z. Čonka, V. Savkiv, and R. Mykhailyshyn,
‘‘Progress and challenges in smart grids: Distributed generation, smart metering,
energy storage and smart loads,’’ Iranian J. Sci. Technol., Trans. Electr. Eng., vol. 44,
no. 4, pp. 1319–1333, Dec. 2020. [11] M. Jaganmohan. (Mar. 3, 2022). Global Smart
Grid Market Size by Region 2017–2023. [Online]. Available:
https://www.statista.com/statistics/246154/global-smart-grid-marketsize-by-region/
[12] Z. Zheng, Y. Yang, X. Niu, H.-N. Dai, and Y. Zhou. (Sep. 30, 2021). Electricity
Theft Detection, [Online]. Available: https://github.com/henryRDlab/
ElectricityTheftDetection
[13] D. O. Dike, U. A. Obiora, E. C. Nwokorie, and B. C. Dike, ‘‘Minimizing
household electricity theft in Nigeria using GSM based prepaid meter,’’ Amer. J. Eng.
Res., vol. 4, no. 1, pp. 59–69, 2015.
[14] P. Dhokane, M. Sanap, P. Anpat, J. Ghuge, and P. Talole, ‘‘Power theft detection
& initimate energy meter information through SMS with auto power cut off,’’ Int. J.
Current Res. Embedded Syst. VLSI Technol., vol. 2, no. 1, pp. 1–8, 2017.
[15] S. B. Yousaf, M. Jamil, M. Z. U. Rehman, A. Hassan, and S. O. G. Syed,
‘‘Prototype development to detect electric theft using PIC18F452 microcontroller,’’
Indian J. Sci. Technol., vol. 9, no. 46, pp. 1–5, Dec. 2016.

[25] 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.
[26] 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.
[27] A. Aldegheishem, M. Anwar, N. Javaid, N. Alrajeh, M. Shafiq, and H. Ahmed,
‘‘Towards sustainable energy efficiency with intelligent electricity theft detection in
smart grids emphasising enhanced neural networks,’’ IEEE Access, vol. 9, pp. 25036–
25061, 2021. [28] K. Phil, MATLAB Deep Learning: With Machine Learning, Neural
Networks and Artificial Intelligence. Seoul, South Korea: Apress, 2017.
[29] 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.
[30] M. Chen, U. Challita, W. Saad, C. Yin, and M. Debbah, ‘‘Artificial neural
networks-based machine learning for wireless networks: A tutorial,’’ 2017,
arXiv:1710.02913.

12. Base papers (IEEE Transactions or good reputed journals)

Leloko J. Lepolesa, Shamin Achari, And Ling Cheng, “Electricity Theft Detection in
Smart Grids Based on Deep Neural Network”, IEEE Access (Volume: 10), 2022.