Reading an article Damage Propagation Modelling for Aircraft Engine Run-to-Failure Simulation and trying to reformat it into a fault detection pipeline of a jet engine, with the help of C-MAPPS dataset and reccurent neural networks.
Download the project from GitHub and change your current directory:
$ (base) cd c-mapps
Use a virtual environment to isolate your environment, and install the required dependencies.
$ (base) python3 -m venv venv
$ (base) source venv/bin/activate
$ (venv) pip3 install -r requirements.txt
To start Fault Detector app, simply write:
$ (venv) python3 app/visualisation.py
For demonstration purposes, the model currently predicts a fault vector. In a real-world scenario, the model would likely predict a fault scalar.
This scalar would represent the overall fault classification (e.g., "OK", "Warning", "Fault") based on the current measurements.
- Reformat the RUL problem to a fault detection (0 - ok, 1 - warning, 2-fault)
- Train a sufficient lightweight models
- Implement a Dash app that allows model inference and helps understand the problem
- Finetune hyperparameters or architecture for each JET engine
- Evaluate models predictions (Residuals)