Socioeconomic Variables Influences Estimation on Non-Technical Losses Results of Low Voltage Distribution Systems

EasyChair Preprint 14599

Abstract

Non-technical losses in distribution systems be represented by the consumed energy amount without billing, mainly due to energy theft and connection frauds. These losses affect the entire utility economic chain, also impacting the number of technical losses. Thus, actions such as field inspections are important to mitigate the problem. Analyzing non-technical losses in low voltage consumer contexts, mostly residential units, relationships between socioeconomic aspects and losses are observed, due to the heterogeneity of the population's living conditions, mainly in developing countries like Brazil. Due to advances in data processing techniques and the current volume of information, the influence of these exogenous aspects can be included in loss expectation models, improving the assertiveness of inspection planning, for example. Thus, this paper seeks to present the relationships between non-technical loss results and socioeconomic information, through statistical analysis and the use of Mutual Information applied to non-linear relationship feature selection. The study will be developed to the demographic units that compose a capital city from the south region of Brazil, however, being a model easily replied to other regions. The results showed greater relationships between non-technical losses and aspects of population income and literacy.

Keyphrases: Análise Espacial, Aspectos Socioeconômicos, Consumidores de Baixa Tensão, Perdas não Técnicas, Seleção de Atributos, informação mútua

@booklet{EasyChair:14599,
  author    = {Leonardo Nogueira F. da Silva and Natalia S. Bastos and Vinicius J. Garcia and Kamila Stromm and Antônio Kaminski Jr. and Daniel P. Bernardon and Alzenira R. Abaide and Otacílio Carneiro Filho},
  title     = {Socioeconomic Variables Influences Estimation on Non-Technical Losses Results of Low Voltage Distribution Systems},
  howpublished = {EasyChair Preprint 14599},
  year      = {EasyChair, 2024}}