Knowlegde extraction from Smart Meters for consumer classification

Abstract

The behavior of an electric customer can be identified through a load profile corresponding to a certain period of time. Information can be available depending on type of customer and customer classification is based only the billing data. But, there are many countries where the Smart Metering Systems are becoming more common and the customer classification and load profiling could be done based on the real consumption data. In paper, is proposed a decision trees based approach for consumer classification in representative categories characterized by typical load profiles using information provided by Smart Meters. For determination of the consumption categories, every customer is characterized by the following primary information: daily (monthly) energy consumption, minimum and maximum loads. The proposed approach was tested using small consumers that don't have a smart metering system implemented. The obtained results it demonstrated that the proposed approach can be used with the success in the optimal operation and planning of distribution systems.