Classificador de Cargas Elétricas Residenciais com Acionamento Individual e Simultâneo

This paper presents an accurate method for identifying residential loads non-intrusively. The performance is analyzed from a dataset with five running loads corresponding to five classes. Higher-Order Statistics and Support Vector Machine (SVM) perform feature extraction and classification, respectively. The results showed that the models completed satisfactorily, and the performance of the work parameter extraction method using Higher Order Statistics can be highlighted. The results obtained in this study were considered excellent, presenting a 100% hit rate for most of the analyzed data. However, it was observed that for class C2 - LED TV, the classifier performance presented slightly lower rates, but still considered excellent.