Dear all,
My lab has been working on an unsupervised NILM algorithm for a while now. We now have a prototype that is described in two conference paper that will be presented in the near future. If interested please have a read. Here are the details of the two papers:
Universal Non-Intrusive Load Monitoring (UNILM) Using Filter Pipelines, Probabilistic Knapsack, and Labelled Partition Maps
Abstract: Being able to track appliances energy usage without the need of sensors can help occupants reduce their energy consumption to help save the environment all while saving money. Non-intrusive load monitoring (NILM) tries to do just that. One of the hardest problems NILM faces is the ability to run unsupervised -- discovering appliances without prior knowledge -- and to run independent of the differences in appliance mixes and operational characteristics found in various countries and regions. We propose a solution that can do this with the use of an advanced filter pipeline to preprocess the data, a Gaussian appliance model with a probabilistic knapsack algorithm to disaggregate the aggregate smart meter signal, and partition maps to label which appliances were found and how much energy they use no matter the country/region. Experimental results show that relatively complex appliance signals can be tracked accounting for 93.7% of the total aggregate energy consumed.
Conference: The 11th IEEE PES Asia-Pacific Power and Energy Engineering Conference 2019 (APPEEC) will be held at The Parisian Macao, Cotai Central, Macao, China on December 1-4 2019 (
https://www.ieee-appeec-2019.org)
Abstract: Non-intrusive load monitoring (NILM) is a research field focused on developing algorithms that can accurately track constituent electrical loads in a system using only the aggregate signal alone (i.e., smart meter). It is widely understood that having a clean signal free of noise and transient behaviour, whether for event-based or state-based methods, can lead to more accurate solutions that will eventually solve the NILM problem. We propose a fast and highly reliable method for producing a block-like representation of signals. Using the same data and disaggregation technique, we compare our algorithm with a recent similar effort and show significant improvements in accuracy (98% vs. 94% tracked energy over three appliances) and run-time (143ms vs. 891s). Application of our method to raw mains power data shows it can generalize to more complex cases.
Conference: The Eleventh Conference on Innovative Smart Grid Technologies (ISGT 2020) will be held at the Grand Hyatt Washington, Washington D.C., USA, February 17-20, 2020 (
https://ieee-isgt.org)
If you have any questions please send me an email to the email address below.
Best,
Stephen.
———
Dr. Stephen Makonin (馬駿豪), PhD, PEng, smIEEE
Adjunct Professor / 研究教授
Engineering Science, Simon Fraser University