How harigittihub/nasa-turbofan-predictive_maintenance works
This repository is a NASA CMAPSS turbofan Remaining Useful Life prediction project, with a Streamlit interface as its user-facing workflow: upload engine data, inspect and select an engine, view a sensor trend, then request an LSTM prediction. The README also documents an offline data-preparation and model-comparison pipeline, including XGBoost and LSTM. The tree and excerpts do not include training code or model artifacts; those documented stages are represented at the existing src directory, and their links rely on README descriptions rather than inspected implementation.
Main entry points: src RUL clipping · streamlit_app.py Selected engine · predict.py RUL inference · load_data.py Raw data loading
Components 19
Data preparation
- CMAPSS data
- Raw data loading
src/data/load_data.py - RUL labels
src/data/make_dataset.py - Exploratory analysis
src - Sensor selection
src - Sliding windows
src - Feature scaling
src - RUL clipping
src
Model development
- XGBoost model
src - LSTM model
src/inference/predict.py - Model artifacts
Interactive inference
- Streamlit interface
app/streamlit_app.py - Uploaded data
app/streamlit_app.py - Selected engine
app/streamlit_app.py - Sensor visualization
app/streamlit_app.py - RUL inference
src/inference/predict.py - Predicted RUL
app/streamlit_app.py - Performance comparison
app/streamlit_app.py
Other components
- User
Connections 19
- User → Streamlit interface: uses
README.md - CMAPSS data → Raw data loading: loads
src/data/load_data.py - Raw data loading → RUL labels: engineers RUL
README.md - RUL labels → Exploratory analysis: analyzes
README.md - Exploratory analysis → Sensor selection: identifies sensors
README.md - Sensor selection → Sliding windows: selects features
README.md - Sliding windows → Feature scaling: creates samples for
README.md - Feature scaling → RUL clipping: prepares features for
README.md - RUL clipping → XGBoost model: trains
README.md - RUL clipping → LSTM model: trains
README.md - User → Uploaded data: uploads
app/streamlit_app.py - Uploaded data → Selected engine: selects engine
app/streamlit_app.py - Selected engine → Sensor visualization: plots sensors
app/streamlit_app.py - Selected engine → RUL inference: requests prediction
app/streamlit_app.py - Model artifacts → RUL inference: loads for inference
src/inference/predict.py - RUL inference → Predicted RUL: returns prediction
app/streamlit_app.py - User → Performance comparison: views metrics
app/streamlit_app.py - XGBoost model → Performance comparison: compared in
README.md - LSTM model → Performance comparison: compared in
README.md