Welcome to TFDA!

This the the repo for Transient Fault Detection algorithm.

This repo is currently used to share and update codes for TFDA.
New algorithms developed in the future will also be uploaded and updated here.

Installing packages

NOTE: The installation instruction below assume that you have python installed on your machine and are using conda as your package/environment manager.

1. Create a new environment: conda create -n oedi python=3.8
2. Activate environment: conda activate oedi
3. Install packages listed in requirements.txt by running the following lines:
   conda install --yes --file requirements.txt
   pip install pip install comtrade
   pip install pip install -U scikit-learn
   pip install scikit-plot
   

Code scipts

Auxiliary functions are in functions.py
fft_3faults.py loads the data in folder Faults and train the model to detect non-fault condidtion or three different fault conditions.
Example of training results -- confusion matrix.

1. 100% accuracy on non-fault conditions.
2. 99% accuracy on type 1 fault condition.
3. 94% accuracy on type 2 fault condition.
4. 100% accuracy on type 3 fault condition.

Example of training results -- ROC curve.