Skip to content

Latest commit

 

History

History
54 lines (29 loc) · 1.65 KB

README.md

File metadata and controls

54 lines (29 loc) · 1.65 KB

Critical Heat Flux Prediction 🔥🔥💣

From Inception ➡️ to Deployment ⬅️ using PyTorch


Business Application for Nuclear Reactors

Flux means flow, critical heat flux is that limit beyond which heat can no longer flow from the solid to the liquid. At this point, vapour starts getting formed. Operating the system beyond this limit leads to overheating and failure.

Hence predicting this value becomes important for the safety of workers in nuclear plants.


What does the code contain

  1. Data Preprocessing and EDA
  2. Torch Dataset and DataLoader preparation
  3. Torch Custom Model building and evaluation
  4. Optuna hyperparameter tuning on model architecture and optimizer's learning rate
  5. Optuna visualizations
  6. Torching saving and loading of model state (incl. weights and biases)

Application UI 👥

Streamlit Deployment Link: https://fluxprediction-6elvtzngufrqfggzt77qst.streamlit.app/

Note: Values may change by a small limit for the same feature values, as the data batches are shuffled each time they are loaded into the model.

CHF1

Explainable AI 📈📉📊

image
image

Next Steps 📃☑️✅

  • Log results on MLFlow
  • Showcase the results on DagsHub
  • Deploy the application on AWS EC2