Detection of fraud or measurement error in big electric power s consumers using Rough Sets based in historical data and in data based in real time / Detecção de fraude ou erro de medição em grandes consumidores de energia elétrica utilizando Rough Sets baseado em dados históricos e em dados em tempo real

AUTOR(ES)

Cristian Mara Mazzini Medeiros Patricio

DATA DE PUBLICAÇÃO

2005

RESUMO

The general purpose of this work is to present a methodology that defines daily behavior profiles of consumers unities of electric energy, fed through high voltage distribution lines, with the purpose of detecting frauds or measurement errors. This methodology allowed the development of a rule based system that uses consumer s static information, i.e., information that does not vary in time and dynamic data, obtained in real time. The development of this system follows the general principles of knowledge discovery in data bases, using the theory of Rough Sets for mining of information, selection of relevant attributes and generation of rules. The system classifies the behavior of the unities, through daily or weekly profiles, in normal or abnormal. The consumer classified as abnormal undergo in-site inspection. The results were considered satisfactory since the rate of correct identification was 64.70%. Therefore, the developed methodology is shown to be capable aid to resolve the problems of commercial losses related to fraud or error of measurement from the utility companies.

ASSUNTO(S)

engenharia eletrica electric power energia elétrica rough sets

Documentos Relacionados

• DATA MINING WITH ROUGH SETS TECHNIQUES
• Detecção de Fraudes em Unidades Consumidoras de Energia Elétrica Usando Rough Sets
• Modelo de avaliação do risco de inadimplência de consumidores de energia elétrica
• Conservação de energia elétrica: um enfoque em consumidores residenciais
• Data mining in large sets of complex data