Electricity Theft Detection Using Machine Learning in Traditional Meter Postpaid Residential Customers: A Case Study on State Electricity Company (PLN) Indonesia
Electricity theft is a major challenge for PT PLN (Persero), particularly in managing 27 million postpaid customers, most of whom still use traditional meters. Detecting and addressing electricity theft has become increasingly complex, requiring more efficient approaches. Unlike smart meters, tradit...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10830511/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Electricity theft is a major challenge for PT PLN (Persero), particularly in managing 27 million postpaid customers, most of whom still use traditional meters. Detecting and addressing electricity theft has become increasingly complex, requiring more efficient approaches. Unlike smart meters, traditional meters lack communication capabilities, making detection reliant on manual processes. This research develops a machine learning model to optimize the Target Operation (TO) process. TO is a list of customers targeted for on-site verification due to suspected electricity theft. This study focuses on optimizing the formation of TO by analyzing monthly electricity usage, particularly in the 450 VA household segment receiving government subsidies. The model aims to reduce reliance on subjective manual observations while ensuring proper subsidy allocation. Various classification models, including Decision Tree, Naive Bayes, Random Forest, K-Nearest Neighbors, Logistic Regression, and Deep Neural Network, were evaluated, with Random Forest achieving the best performance across simulations. A sequential evaluation method is introduced to enhance accuracy through layered filtering, where detection results from the three-theft model are further filtered using the two-theft and one-theft models, resulting in a more precise TO. The combination of Random Forest and K-Nearest Neighbors achieved the highest performance, with an accuracy of 0.89, precision of 0.83, recall of 0.98, F1-Score of 0.90, and AUC of 0.89. These findings demonstrate the model’s effectiveness in delivering reliable TO recommendations, supporting PLN’s operational strategies, and offering practical benefits through a more objective, standardized TO process that minimizes human error and improves efficiency. |
|---|---|
| ISSN: | 2169-3536 |