Commercial Losses, Energy Theft, Non-Technical Losses, Power Distribution System, Soft Computing Techniques
Abstract
Non-technical losses, or commercial losses, result from various factors, such as energy theft via irregular or illegal connections and energy meter fraud. These events cause significant financial damage to the power grid, power utilities, and consumer units. These losses impact the financial results of power utilities and damage the power distribution systems' reliability and efficiency. Therefore, in this study, a comparative analysis is made among classifiers based on soft computing techniques, namely support vector machine and multilayer perceptron, fuzzy ARTMAP neural networks for detecting non-technical losses. The classifiers’ performance is evaluated based on the confusion matrix metrics. Results show that the multilayer perceptron performs better with accuracy, precision, recall, and F1-Score of 98.15%, 88.67%, 91.19%, and 89.91%, respectively.
