This is a repository that contains some legacy code (mainly tensorflow 2) that corresponds to different experiments performed for remaining useful life prediction.
The code is not supported and it is recommended that future users try to implement the techniques described from scratch. It is only provided as reference, since there was some interest in accessing the code.
In this repository, there are python files that implement a simple GNN library, functions/scripts/notebooks that are meant to execute experiments, and notebooks that are meant to execute and visualize intermitent results of experiments. There are also some notebooks that are not meant to be read (or understood) and it may be easy to understand which are the ones by simple inspection.
This is to clarify that the bulk of the code and scripts in the repository is meant to be ignored. There are some notebooks that may be of use to researchers which I list in what follows:
- LSTM+tCNN -based model on FEMTO bearings dataset
- Causal GNN-based model on Fictitious dataset
- Causal GNN-based model on FEMTO bearings dataset
Do not hesitate to drop me an email for some "behind the scenes" on advice for implementation. However, I cannot promise to reply in a timely manner since I am not working as a researcher anymore.