Electric load classification by binary voltage-current trajectory mapping

Characterization of electric loads provides opportunities to incorporate detailed energy usage information into applications such as protection, efficiency certification, demand response, and energy management. This paper proposes a low computational cost, but yet accurate method, to extract signatures for load classification and characterization. Instead of utilizing digital signal processing and frequency-domain analysis, this paper abstracts the similarity of voltage-current (V-I) trajectories between loads and proposes to map V-I trajectories to a grid of cells with binary values. Graphical signatures can then be extracted for many applications. The proposed method significantly reduces the computational cost compared with existing frequency-domain signature extraction methods. Test results show that an average of over 99% of success rate can be achieved using the proposed signatures.