@article {10.3844/ajeassp.2026.117.132,
article_type = {journal},
title = {Optimization of Non-Technical Electricity Loss Detection Using Artificial Neural Networks: Case of Cameroon Distribution Network},
author = {Bernard, Lekini Nkodo Claude and Samuel, Bell Serge and Valjacques, Nyemb Nsoga and Urbain, Nzotcha},
volume = {19},
year = {2026},
month = {Jul},
pages = {117-132},
doi = {10.3844/ajeassp.2026.117.132},
url = {https://thescipub.com/abstract/ajeassp.2026.117.132},
abstract = {Non-Technical Losses (NTL) represent a critical challenge for power utilities, particularly in developing countries. This paper proposes an unsupervised deep learning approach based on an autoencoder to detect abnormal electricity consumption patterns in the Cameroonian distribution network. The model learns normal customer behavior using historical consumption data and identifies anomalies through reconstruction error analysis. Statistical thresholds are applied to classify customers as normal, suspicious, or fraudulent. Experimental results show that the proposed method achieves high detection reliability while reducing inspection costs. The approach provides a scalable and practical solution for utility companies seeking to enhance revenue protection and energy security.},
journal = {American Journal of Engineering and Applied Sciences},
publisher = {Science Publications}
}