Abstract: Methods and systems for controlling power supplied to a plurality of appliances on a power line. Poly phase power measurements are collected from the power line by a dedicated energy metering chip are sampled at frequencies in the range of 0 kilo samples per second to 32 kilo samples per second and converted to digital power measurements. The digital power measurements are received by a real-time microcontroller. The poly phase power measurements are analyzed in real time, the poly phase power measurements are disaggregated, and a power report regarding a power usage of each appliance is generated by a computer processing unit, CPU. A power report is transmitted to a mobile application, power on/off commands for each appliance are received from the mobile application, and each appliance is powered on/off based on the power on/off commands.

Abstract: An apparatus and method are used for disaggregating and classifying energy consumption data from the overall energy consumption data of a building. Both high frequency sampling data classified by a convolutional neural network running on a cloud server and low frequency sampling data analyzed by a convolutional neural net running locally on an electric smart meter are used in the method. A control circuit switches between high and low frequency sampling modes based on the activity level of the building. Individual devices can also be controlled through a mobile application connected to the control circuit.