Sequencing of actions:
- The turbine generator’s historical data was collected and analyzed
- The relevant parameters characterizing the state of the turbine generator were identified
- The values of stator current variation till failure were analyzed
- A linear regression of the stator winding temperature from the total power of the turbine generator was constructed based on the data for one month of its normal operation
- and actual and calculated temperatures were compared across the entire sample of values provided
Results
- A systematic excess of the actual temperature of the stator winding over the calculated temperature in specific grooves and temperature peaks were detected.
- Actual temperature of stator winding
- Calculated temperature of stator winding
The charts of the change in the calculated and actual temperatures of the stator winding based on the example of two grooves
- The predictive analysis model thus developed can detect abnormal temperatures in the stator winding of a turbine generator 1.5 years before failure.
- The model can predict the onset of failure in other grooves well in advance
