Bearing fault detection from vibration signals

The brief

A plant's motors and pumps fail without warning when bearings wear out. Using vibration recordings, build a classifier for healthy bearings and inner-race, outer-race and ball faults, so maintenance can be planned. Good looks like: high accuracy across fault types and loads, with features an engineer can relate to the physics.

What you will build

1. Read the vibration recordings and plot them
2. Extract time and frequency features (RMS, kurtosis, FFT bands)
3. Train and compare classifiers
4. Test on a motor load not used in training
5. Explain which features point to which fault

Free, open data to build it with

• CWRU Bearing Data Center ↗ Case Western Reserve University

Check each dataset's licence before you publish your work. Your mentor gives your team its own version of the brief (data, scope or region), so every team's project is its own.