Visible to the public "Single-channel compressive sampling of electrical data for non-intrusive load monitoring"Conflict Detection Enabled

Title"Single-channel compressive sampling of electrical data for non-intrusive load monitoring"
Publication TypeConference Paper
Year of Publication2015
AuthorsM. Clark, L. Lampe
Conference Name2015 IEEE Global Conference on Signal and Information Processing (GlobalSIP)
Date PublishedDec. 2015
ISBN Number978-1-4799-7591-4
Accession Number15807748
KeywordsBuildings, compressed sensing, Current measurement, Demodulation, electrical data, electrical signal, Frequency measurement, high frequency electrical measurement, Home appliances, Monitoring, NILM, nonintrusive load monitoring, power system measurement, pubcrawl170104, random demodulation, random filtering, sampling channel, sampling methods, sampling rate, single-channel compressive sampling

Non-intrusive load monitoring (NILM) extracts information about how energy is being used in a building from electricity measurements collected at a single location. Obtaining measurements at only one location is attractive because it is inexpensive and convenient, but it can result in large amounts of data from high frequency electrical measurements. Different ways to compress or selectively measure this data are therefore required for practical implementations of NILM. We explore the use of random filtering and random demodulation, techniques that are closely related to compressed sensing, to offer a computationally simple way of compressing the electrical data. We show how these techniques can allow one to reduce the sampling rate of the electricity measurements, while requiring only one sampling channel and allowing accurate NILM performance. Our tests are performed using real measurements of electrical signals from a public data set, thus demonstrating their effectiveness on real appliances and allowing for reproducibility and comparison with other data management strategies for NILM.

Citation Key7418305