Download the REDD data set
In the previous note, we downloaded and built the NILMTK environment.
( NILM (non-intrusive power load monitoring) study notes-preparations (1) configuration environment NILMTK Toolkit )
In this chapter, we will prepare a dataset for training and testing.
This data is provided by the MTK team for free. You need to log in to their website, email them to them, get the account and password, and download it.
Resource address: http://redd.csail.mit.edu/
Send an email at random and ask for the REDD data set in one sentence, and the account and password will be returned (send an email to [email protected] and the machine will automatically reply to your account and password)
so my account password is not provided.
Providing data can be seen after two high and a low frequency
herein NILMTK using low frequency data processing, so the download low_freq.tar.bz2
After downloading and decompressing, you can see the data of 6 houses.
Copy the low_freq folder to the project directory.
The original data downloaded is in .dat format. NILMTK does not support this format, so you need to convert the data format.
Create a new python file REDDToH5.py, the conversion method directly calls the method in the library is very simple, the parameter is the path of the original data low_freq, and the output path is just fine.
code show as below:
from nilmtk.dataset_converters import convert_redd
convert_redd(r'F:\MyProject\RNN\NILMTK\low_freq',r'F:\MyProject\RNN\NILMTK\low_freq\redd_low.h5')
After running this script, you can see that there is one more .h5 file in the output path. This is what NILMTK can directly use.
At this point, our data set is also ready.
The next step is to use the data set, and the NILM library to decompose and predict.
How to use it is detailed in my article later.
NILM (non-intrusive power load monitoring) study notes-use NILMTK Toolkit, REDD data set, CO and FHMM two algorithms
https://blog.csdn.net/wwb1990/article/details/103904340