| Abstrakt: |
Non-Technical Losses (NTLs) have become a significant concern for many countries and often constitute the total electricity distributed. Utilities around the world incur huge costs due to NTLs estimated at billions of dollars. Therefore, reducing NTLs has become pertinent for utilities to improve revenue generation, profitability on investment, and grid reliability. In the last decade, artificial intelligence and machine learning have been the predominant trend to detect NTL in the industry and academia to elicit scrutiny of suspicious customers. While these methods have succeeded in identifying and detecting NTLs, it continues to thrive, particularly in developing economies. Therefore, it is crucial to investigate the issues hindering customers from paying for electricity consumed and the quantum effect on the utilities. Bayesian Network, a machine learning approach that has proven to help depict causes and effects, has been explored to provide valuable insights into the impact of some of the significant causes of NTLs. We evaluated the resultant Bayesian model with five other Machine Learning models. Causes of NTLs like unpaid bills and electricity theft have been a severe challenge to utilities from inception. Thus, utilities are facing enormous business challenges and are seeking some form of transformation to ensure sustainability. The study shows that revenue collection through customers paying bills is fundamental to this strategy. |
Full text from SpringerLink