摘要 |
<p>Failure of an appliance is predicted by accessing multiple days of aggregate power use data for a home, searching for peaks in the power use data that occurred during different days in the multiple days, determining that a peak during a later day of the multiple days has a larger magnitude than a peak during an earlier day, and identifying an appliance responsible for the larger magnitude of the peak during the later day. A user interface is then sent to a user device to indicate that the identified appliance is failing.</p> |