Table 6 Comparison of power disaggregation accuracy values (EACC) for recently proposed NILM methodologies. The reported performance is the average EACC across houses 1, 2, 3, 4, and 6
From: Robust energy disaggregation using appliance-specific temporal contextual information
NILM method | Year | Dataset | EACC (%) |
|---|---|---|---|
General sparse coding [69] | 2010 | REDD-1/2/3/4/6 | 56.4 |
Discriminating sparse coding [69] | 2010 | REDD-1/2/3/4/6 | 59.3 |
Temporal ML [70] | 2011 | REDD-1/2/3/4/6 | 53.3 |
Powerlets-PED [67] | 2015 | REDD-1/2/3/4/6 | 72.0 |
Greedy deep sparse coding [71] | 2017 | REDD-1/2/3/4/6 | 62.6 |
Exact deep sparse coding [71] | 2017 | REDD-1/2/3/4/6 | 66.1 |
Supervised GSP* [68] | 2018 | REDD-1/2/3/4/6 | 67.8 |
Unsupervised GSP* [68] | 2018 | REDD-1/2/3/4/6 | 74.6 |
Proposed TCI method | 2020 | REDD-1/2/3/4/6 | 76.3 |