Can domain knowledge help to model a battery system?
In order to get this project up and running on your machine there are several steps to be taken:
In order to be able to run the provided jupyter notebooks there are several dependencies needed to be installed. Using an anaconda shell in the Battery-System folder, please enter :
conda env create -f environment.yml
This command will install all necessary dependencies needed for this project in an environment called "battery-system". Activate this environment using:
conda activate battery-system
Execute the following command inside the activated environment to create a new kernel:
python -m ipykernel install --user --name battery-system --display-name "battery-system"
Now the installation should be completeted and all necessary dependencies should be included.
The data used for analysis needs to be included separately
After extracting the zip, you should be provided with a folder named fobss_data
Battery-System/data/raw/Now you are all set! Navigate to Battery-System/notebooks/data/fobss_overview.ipynb to get an overview of the data.
The data we used in our approach is the FOBSS (Frequent Observations from a Battery System with Subunits) dataset. The data was conducted on a battery system with several battery packs each monitored by a subunit of the Battery Management System (BMS). They monitored the current, voltage and temperature during several charge, discharge and rest procedures, called profiles.
The measurements were taken on three different levels of the system: