• Abstract: Energy disaggregation, also known as non-intrusive load monitoring (NILM), aims to separate the aggregated electricity consumption of a power system as a whole into individual functional components being used. Disaggregating the meter readings of a residential house can provide occupants with energy consumption of individual appliances. Such disaggregated feedback can efficiently motivate occupants' interests to their daily power usage patterns and make incredible contributions to energy conservation. In this project, we develop consumer-oriented, easy-to-use and scalable approaches to energy disaggregation, using readily-available and easily-available appliances knowledge.
• Related Publications: [SmartGridComm 2014], [NILM Workshop 2014], [ICDM Demo 2014], [TPDS 2015].
• Demo Video: [SmartSaver@YouTube] or [SmartSaver@youku(优酷)].

• Abstract: Accurate load curve data is important on the energy demand side (e.g., residents) as well as the supply side (e.g., power companies), while the data are subject to corruption caused by many factors, such as communication failures, meter malfunctions, unexpected interruption or shutdown in electricity use, unscheduled maintenance, and temporary close of production lines. Therefore, load curve data cleansing has caught more and more attention recently. Our work focuses on i) outlier detection to aggregated load curve from hundreds or thousands of households and ii) corrupted data identification to the smart meter data from individual households.
• Related Publications: [TSG 2014], [CIKM 2014].

• Abstract: Occupancy detection, also termed as presence detection or occupancy monitoring, aims to determine whether or not someone is in a room or house. For a long time, it has been recognized as a simple yet efficient way to help save building energy. For example, with accurate occupancy information of a building, automation systems can effectively control its heating, ventilation and cooling (HVAC) and lighting system in real-time to reduce the energy waste and improve the building energy efficiency. Our work focuses on non-intrusive occupancy detection (NIOD) via the load curve data or smart meter data.
• Related Publications: [SmartGridComm 2015].

• Abstract: Power monitoring is critical to the efficient operation and energy saving of datacenters. Fined-grained power monitoring, however, is extremely challenging in legacy datacenters that host server systems not equipped with power monitoring sensors. Installing power monitoring hardware at the server level not only incurs high costs but also complicates the maintenance of high-density server clusters and enclosures. In this project, we have developed a purely software-based solution to this challenging problem. We implement and evaluate our solution over a real-world datacenter with 326 nodes, and the results show that our solution can provide high precision power estimation at both rack and server levels.
• Related Publications: [Middleware 2015].

• A complete dataset for non-intrusive occupancy detection (NIOD) via load curve data. [download]
• Two complete datasets for energy disaggregation or load curve data cleansing. [download]

© G. Tang, last updated: Aug. 12, 2015.