Research Article
Nonintrusive Load Monitoring Based on Advanced Deep Learning and Novel Signature
Table 1
Public datasets for NILM [
31].
| | Dataset | Number of houses | Duration per house | Appliance sample frequency | Aggregate sample frequency |
| | REDD (2011) | 6 | 3–19 days | 3 sec | 1 sec, 15 kHz | | BLUED (2012) | 1 | 8 days | N/A | 12 kHz | | Smart (2012) | 3 | 3 months | 1 sec | 1 sec | | Tracebase(2012) | N/A | N/A | 1–10 sec | N/A | | Sample (2013) | 10 | 7 days | 1 min | 1 min | | HES (2013) | 251 | 1 or 12 months | 2 or 10 min | 2 or 10 min | | AMPds (2013) | 1 | 1 year | 1 min | 1 min | | iAWE (2013) | 1 | 73 days | 1 or 6 sec | 1 sec | | UK-DALE (2016) | 5 | 1–3.5 years | 6 sec or 16 kHz | 1 or 6 sec, 16 kHz |
|
|
