A novel approach for home appliances identification using Vector Projection Length and Stockwell Transform

摘要

Identification of residential electrical appliances has been a highlight area of research in recent years due to increasing energy consumption in the residential sector around the world. This paper addressed a novel method for recognition of electrical appliances. Thus, it is proposed to apply the Stockwell Transform, a time-frequency domain transform, into current data collected from 17 commonly present electrical appliances in our houses. With these data, we applied an image pattern recognition technique, called Vector Projection Length. In this case, the idea is to recognize patterns in the appliances transformed into time-frequency domain. An extensive analysis technique simulation is presented here, showing the variation of the technical parameters and their influence on the results of identification.