Home / Industrial & Manufacturing / Predictive Maintenance
IndustrialOEE — AvailabilityLSTM

Predictive Maintenance — catch the failure before it happens

Remaining Useful Life (RUL) prediction for industrial turbine engines, comparing an LSTM sequence model against XGBoost on the NASA CMAPSS FD001 benchmark.

⚠️ This is an independent portfolio project built on public data — not a paid client engagement. Results are shown as-is, including known limitations. The $10.8M ROI figure is a self-modeled estimate based on public industry cost benchmarks, not a measured client result.

Key results

ModelLSTM (PyTorch) — beat XGBoost
RMSE17.08
Estimated cost avoided per failure event~$450K
Live dashboardOpen live demo →

How it works

Sensor data (temperature, pressure, vibration proxies) from the NASA CMAPSS benchmark feeds a sequence model that learns degradation patterns over time and predicts how many operating cycles remain before failure. A two-tab Streamlit dashboard lets an operator explore predictions and estimated savings by fleet size.

Have a similar problem in your operation?
Talk to us about your project