Active sensing approach to electrical load classification by smart plug

Classification of electrical loads presents multiple benefits. From the ability to inform customer how their appliances contribute to the overall electricity cost, to allowing electric companies and virtual aggregators to institute demand response mechanisms. In this paper we propose a new type of active sensing method which breaks down real-time data into fixed segments which can be quickly analyzed and used to classify electrical loads to help reduce peak power demand. We have designed and implemented a simple smart plug with a bidirectional triode thyristor, gathered data from household appliances and performed classification. Our results show high accuracy with significantly shorter times than other similar proposed methods.