BLOND DATA CONVERTER

Usage

Downloading the BLOND-250 dataset

run

python download_sum.py

This will download all summary files for each day for each MEDAL and CLEAR unit. The folder structure will be kept as it is on the FTP.

Converting BLOND metadata to NILMTK

run

cd metadata_converter/ && python convert_blond.py && cd ..

This will build the building.yaml file and copy the dataset.yml and meter_devices.yml to the dist/ folder. This dist folder containing these three files will be needed for the data conversion.

Converting BLOND to NILMTK

run

cd data_converter/ && python convert_sum.py && cd ..

This will convert BLOND to a NILMTK format. The result will be written as converted_sum.hdf5 in the data/ folder.

Running tests

The notebooks/ folder contains a IPythonNotebook computing some basic statistics about the BLOND dataset.

If you want to test NILMTK disaggregation models you can run the test_top_k.py script which will disaggregate for k=5 appliances. You can change the parameters in the file in order to get different results. The script can be run with