HIFDA - High-Frequency Electrical Signals from Household Appliances Dataset
Creators
Description
This work aims to provide a new dataset containing high-frequency steady-state electrical signals from individual common household appliances. A total of 14 appliances were captured with an acquisition rate of 100 kSPS, capturing only the times when the target appliance was active and consuming power (no idle states). Also, the empty grid (no active appliance) was captured in the same conditions.
For the creation of this dataset, 50 windows of the voltage and current signals were taken from each of the 14 isolated devices and the empty grid, each window containing approximately 54000 samples acquired at 100 kSPS, implying windows of 5.4 seconds. For convenience, the windows were split again to create three additional datasets, containing time slots of 10.24, 163.84 and 1310.72 milliseconds, in order to test different architectures when used for training the target neural network. The size of these windows has been set so that the number of data contained in each window is a power-of-two figure, seeking to create windows containing at least time slots comprising 10 ms, 100 ms and 1000 ms at 100 kSPS, respectively. This decision was made to increase the performance of the processing device that uses this data for its intended purpose.
The dataset provided contains five main folders. One of this folders, "0.Img_Appliances", contains pictures of the 14 appliances whose electrical signals were captured to create this dataset. The rest of the folders correspond to the different window divisions, as explained before. Inside each of them there is a "Current" and a "Voltage" folder where, inside each, there is a folder for every appliance that was captured, containing the data in multiple files with text (.txt) format. It is important to keep in mind that, even though the file quantity is greater on the smaller window datasets, the information that all of the datasets contain is mostly the same, as it all comes from the full time records dataset. The bandwidth of the captured voltage signal ranges from 300 Hz to 50 kHz, so the fundamental component of the grid, located at 50 Hz, does not appear on the captured data. The captured current signal, on the other hand, has a bandwidth ranging from 30 Hz to 50 kHz approximately.
The appliances included are:
- Air conditioner
- Charger
- Computer
- Hair dryer
- Empty grid
- Griddle
- Heater
- Iron
- Coffee maker
- Laptop
- Light
- Microwave
- Monitor
- Vacuum
- Washing machine
Files
HIFDA_HF_electrical_signals_dataset.zip
Files
(4.4 GB)
| Name | Size | Download all |
|---|---|---|
|
md5:259a476892da18b53c516a93c9129322
|
4.4 GB | Preview Download |
Additional details
Related works
- Is derived from
- Poster: 10.1109/I2MTC60896.2024.10560829 (DOI)
- Is described by
- Journal article: 10.1038/s41597-025-04859-3 (DOI)
References
- Kahl, M., Ul Haq, A., Kriechbaumer, T. & Jacobsen, H. A. Whited - a worldwide household and industry transient energy data set. Int. Work. on Non-Intrusive Load Monit. http://nilmworkshop.org/2016/proceedings/Poster_ID18.pdf (2016).
- Kriechbaumer, T. & Jacobsen, H. A. Blond, a building-level office environment dataset of typical electrical appliances. Sci. Data 5, 180048, https://doi.org/10.1038/sdata.2018.48 (2018)
- Kelly, J. & Knottenbelt, W. The uk-dale dataset, domestic appliance-level electricity demand and whole-house demand from five uk homes. Sci. Data 2, 150007, https://doi.org/10.1038/sdata.2015.7 (2015).