Electrical Load Classification with Open-Set Recognition

Full utilization of renewable energy resources is a difficult task due to the constantly changing demand-side load electrical grid. Demand-side management would solve this crucial problem. To enable management, knowledge about composition grid required, as well ability schedule individual loads. There are proposed Smart Plugs presented in literature capable such tasks. The problem, however, that these methods lack detect if previously unseen connected. Misclassification loads presents problem for estimation and scheduling. Open-set recognition by providing way samples not belonging any class used during training classifier. This paper evaluates novel application open-set classification. Two approaches were examined, both offer promising results. A Support Vector Machine based approach was first evaluated. second, more robust method modified OpenMax-based algorithm