AI Techniques in Detection of NTLs: A Comprehensive Review

Abstract

In the operation of power grid, worldwide, non-technical losses (NTLs) occur in a massive amount of proportion which is observed up to 40% of the total electric transmission and distribution losses. These dominant losses severely affect to adverse the performance of all the private and public distribution sectors. By rectifying these NTLs, the necessity of establishing new power plants will automatically be cut down. Hence, NTLs have become a critical challenge to do research in this emerging area for researchers of power systems due to the limitations of the current methodologies to detect and fix up these prominent type of losses. The existing survey so for basically contains the detail of identification of non-technical losses by machine and deep learning methods while this paper is a complete trouble shooting to resolve this issue by systematic approach. To address this, causes of NTLs along with its impact on economies and types of NTLs are elaborated in various countries. In addition, we have also prepared a comparative analysis based on several essential parameters. Further, implementation process of detection of NTLs or electricity theft based on Machine Learning or Deep Learning has also been demonstrated. Moreover, major challenges of detection of NTLs or electricity theft based on ML and Deep Learning, and its possible solutions are also described. Hence, definitely this comprehensive survey will help to the leading researchers to reach a new height in this thrust area.