Mini-project
Energy, Power & Utilities
Wind turbine gearbox health from SCADA data
Use turbine control-system data to spot a gearbox running hotter than it should and warn weeks ahead.
The brief
Wind turbines sit in remote places and a gearbox failure means a crane and weeks of lost power. Using normal operating data (wind speed, power, temperatures), model what a healthy gearbox temperature should be, and flag the turbine when the real temperature drifts above it. Good looks like: a normal-behaviour model trained only on healthy periods, a drift alarm with a useful lead time, and a count of false alarms.
What you will build
- Load and clean the turbine data
- Model normal gearbox temperature from wind and power
- Compute the gap between real and expected
- Set an alarm rule and a lead-time target
- Test on a period that ended in a fault
- Write a maintenance note
Free, open data to build it with
Sign up free to get the data and links
It takes a minute with your email. AiTeky Pass (₹999/month or ₹9,999/year) is free for your first 30 days.