A Low-Cost Embedded System for Non-Technical Loss Detection in Low-Voltage Distribution Networks
A low-cost wireless embedded system was developed and evaluated for the local and real-time detection of non-technical losses caused by electrical bypasses in low-voltage service drops. The architecture consists of two ESP32-based measurement nodes equipped with PZEM-004T V3 modules and interconnected through ESP-NOW. Node A measures the electrical variables at the service connection point, whereas Node B acquires the billing-point variables, processes the received telemetry, and displays the operating status through a local web dashboard. The detection model incorporates the estimated resistive loss of the service drop and applies a current-dependent dynamic threshold to distinguish technical losses from unexplained power discrepancies. Measurement pairs are validated through sequence identification, timestamp-based synchronization, and wireless-link quality criteria before classification. The experimental campaign comprised 40 scenarios, including 20 controlled tests and 20 different residential service drops with conductor cross-sectional areas of 6 and 10 mm², lengths from 4 to 25 m, and different load profiles. Thirty-eight acquisitions satisfied the established quality criteria. The system obtained 9 true positives, 28 true negatives, 0 false positives, and 1 false negative, corresponding to an accuracy of 97.37%, sensitivity of 90.00%, specificity of 100.00%, and precision of 100.00%. The ESP-NOW link achieved an overall packet error rate of 0.79% and a mean transmission latency of 4.36 ms. The results show that technical-loss compensation and acquisition-quality control reduce false alerts under the evaluated high-load, transient, and nonlinear operating conditions.
Authors
- Jezzy James Huaman Rojas (ORCID: https://orcid.org/0000-0002-5158-6393)
- Andy Quispe Mendoza
- Cristhian Paolo Poma Mercado
- Yordan Zosimo Cerron Armas
Institutions
- Universidad Continental (PE)
Publication Details
- Journal
- International Journal of Electrical and Electronics Engineering
- Published
- 2026-11-13
- DOI
- https://doi.org/10.14445/23488379/ijeee-v13i10p108
- Primary Topic
- Electricity Theft Detection Techniques
- Type
- article
- Field-Weighted Citation Impact
- 0.00