Automatic Appliance Identification Based on Consumption Time Series and Event-Driven Processing

The deployment of smart meters is increasing in modern societies. A fine-grained metering data acquisition and processing is crucial to help the smart grid stake holders. The classical data sampling approach is time invariant. It includes in the acquisition, transmission, and processing stages a large amount of redundant data. This deficit can be eliminated by employing the event-driven sampling, which provides a realtime data compression. Therefore, a novel event-driven adaptive-rate sampling approach is utilized for the appliances consumption recording and features extraction. The relevant features related to the appliances consumption patterns such as power and current are subsequently utilized for their identification by using the Artificial Neural Network classifier. Results confirm an 8 folds compression gain and the processing effectiveness of the suggested approach while securing 95.1% average classification accuracy.