• Optiputer

  • HP Project

  • Video Conferencing system

  • NMO Lab

  • CenSEPS

  • Storage Systems Research Center (SSRC)

  • Appliance Load Monitoring through Ambient Noise Analysis and Classification (A.L.M.A.N.A.C.)- John Jacobs, M.S. Thesis

    We present a method of estimating electric power consumption in residential settings by identifying when major appliances are in use via a signal classification approach. First, we capture a power consumption profile of each device to be monitored and store it in a database. We then use data collected from a variety of sensors and to train a classifier that attemptes to identify the activation state of all of the monitored devices at any time. Once the classifier has been sufficiently trained, we put it into a detection mode, in which changes in device states are estimated in near-real-time. Combining these detected state changes with the previously-captured device energy profiles, we will be able to construct a detailed time-line view of the aggregate household energy usage, broken down into contributions from individual appliances and devices. Out system makes use of sensor fusion by aggregating signals from an acoustic microphone, an AM radio, and a current sensor; mean test accuracy rate was improved to 91% from 61% when using a current sensor alone.

Subscribe to Information Technology Institute Events (RSS) Subscribe to Information Technology Institute Events (iCal)