A wind energy generation company that generates 810 MW of wind power has approached PREDICT to provide a solution to increase their energy production capability by monitoring the health of wind turbines.
Detect abnormal power conditions
Detect abnormal temperature conditions
Predict’s predictive maintenance solution aimed to increase the capability of these assets by providing valuable information on future failures and system health.
- Improve the operation capability of the asset
A thorough study of the engineering of the wind turbines and associating it with the historic data of the sensors from the wind turbines led to efficient monitoring of the operation of the wind turbine and assessing its remaining useful life.
- Determine abnormal power conditions:
By using predictive analytic algorithms, the turbine characteristics were modelled which helped in determining abnormal conditions in power.
- Determine abnormal temperature conditions:
The temperature of the bearings of the wind turbine was monitored and, abnormal conditions were detected. This was instrumental in predicting when maintenance needs to be performed to prevent serious damage to the wind turbine.
PREDICT analysed the historic data from different sensors for 10 wind turbines. Deviations in the turbine’s performance were detected by modelling the turbine characteristics of each turbine using machine learning techniques. The modelled turbine characteristics were conducive in detecting abnormal power conditions.
Correlating the engineering of the turbine with the historical data, PREDICT could determine the behaviour of speed and temperature in different conditions. This led to the detection of abnormal temperature conditions.
This information provided useful insight for the client into the wind turbine performance characteristics, including how performance deviates between the 2 wind farm locations and identification of how turbines are deviating from their regular performance.