Code to apply the Unscented Kalman Filter (UKF) and the Extended Kalman Filter (EKF) to a measurement from the experimental zero-power reactor CROCUS for pseudo-live reactivity estimation and simultaneous reactor parameter estimation.
Install this project by installing all dependencies and subsequently cloning the repository.
You need a working latex installation. You might also need some or all of the following packages: dvipng texlive-latex-extra texlive-fonts-recommended. Linux example (Windows with a working latex installation is fine, too):
sudo apt-get install texlive dvipng texlive-latex-extra texlive-fonts-recommended
The code runs under Python 3.8.5, as well as Python 3.7.3.final.0. Tests are run automatically with Python 3.8.5 after every push or pull request to master.
The dependencies are listed in requirements.txt (respectively requirements_3-7-3.txt for Python 3.7.3.final.0).
Note: This project depends on pykalman, which is not listed in the channels that Anaconda uses by default. If you use Anaconda, use conda config --append channels conda-forge.
Note: In order to run tests, such as the ones in test_RKKF.py, you also need pytest: pip install -U pytest or conda install pytest.
With Pip:
Python 3.8.5:
pip install -r requirements.txt
Python 3.7.3:
pip install -r requirements_3-7-3.txt
With conda:
Python 3.8.5:
conda config --append channels conda-forge
conda install --file requirements.txt
Python 3.7.3:
conda config --append channels conda-forge
conda install --file requirements_3-7-3.txt
git clone https://github.com/Grim-bot/CROCUS-UKF.git and you're good to go!
Once you have installed dependencies and cloned the repository, simply run the main function of Reactor_Kinetics_Kalman_Filter.py. If you run it in an IPython shell (e.g., using an IDE), the figures will appear in the console. Regardless, all figures are saved to the figures/ subdirectory in pdf format.
python Reactor_Kinetics_Kalman_Filiter.py
pytest -x
Have a look at the Issues! You can also contribute by suggesting a code documentation framework.
The data directory contains measurements from CROCUS, as well as prior estimates of the reactor parameter.
The figures directory is where all figures are saved (using pdf format). Figures should not be committed to git, since they are stored in binary format. If you wish to commit figures, use git-lfs.
The most important executable python scripts are found at the root level of the repository. Namely:
Reactor_Kinetics_Kalman_Filter.py: This file defines and executes a series of simulations using the UKF and EKF and generates plots.
The tests directory contains python files that define tests to be run at every new push (or pull request) to the master branch.