http://lattes.cnpq.br/7750748954265218; SILVA, Harllan Andryê Bezerra.
Resumo:
The consumption of electric energy has been increasing every day. We need to use electric
power in a conscious way, because the natural resources that are used for the generation of energy can end up due to its inefficient use. The population growth of the last decades, the appearance of more electronic devices and appliances generate an excessive consumption of energy. Due to the growth in the consumption of electric energy, it is necessary to implement energy efficiency programs, which are carried out through the introduction of new technologies, an incentive to change the consumer’s habit and rational use of electric energy. The focus of this work is on the residential sector, which is the second largest consumer of electricity in Brazil, and since there are consumers who share similar characteristics and load patterns, this allows the use of data grouping. Thinking about that, the use of clustering to support energy efficiency programs in the analysis of consumer data and in the creation of representative groups of a population is proposed. Groups creation helps the utility to provide commercial offers or specific recommendations for specific groups, reduce the complexity of the analyzes that would have to be done in a population, and get personalized, more effective and equitable relationships between energy suppliers and their customers. The clustering will provide the application of solutions that help the consumer to use electricity efficiently, from the moment he receives information about his consumption and how he can use that information, knowing what they will provide as a result. This work began with the investigation of measures of dissimilarity to represent the similarity between profiles of electric energy consumption (one of the factors used for the clustering) and among the three measures used the Euclidean distance stood out with the best results in the experiments made, either by varying the number of observations of the series or the database. After that, clusters were made using 4 factors extracted from the database and thus 15 clustering scenarios were created from the combination of these factors. Through the results of these clustering it was possible to reduce the number of scenarios to be similar and also to choose the most relevant
scenarios to consider when creating groups of residential consumers.