Prediction of Dry-Low Emission Gas Turbine Operating Range from Emission Concentration Using Semi-Supervised...

Area of Science:

• Mechanical Engineering
• Combustion Science
• Artificial Intelligence

Background:

• Dry-Low Emission (DLE) technology utilizes lean pre-mixed combustion to minimize nitrogen oxides (NOx) and carbon monoxide (CO) emissions.
• Gas turbines face operational challenges like tripping due to frequency deviations and combustion instability, especially under sudden disturbances or poor load planning.

Purpose of the Study:

• To develop a semi-supervised technique for predicting suitable operating ranges in DLE gas turbines.
• To establish a tripping prevention strategy and guide for efficient load planning in gas turbine operations.
• To enhance the reliability and efficiency of DLE gas turbines through advanced control strategies.

Main Methods:

• A hybrid semi-supervised learning model combining Extreme Gradient Boosting and K-Means algorithms was developed.
• The model was trained and validated using actual plant data from DLE gas turbine operations.
• Performance was benchmarked against other machine learning algorithms like decision trees, linear regression, support vector machines, and multilayer perceptrons.

Main Results:

• The proposed model achieved high prediction accuracy for combustion temperature (R²=0.9999), NOx (R²=0.9309), and CO (R²=0.7109).
• It successfully identified optimal and safe operating regions for DLE gas turbines, typically between 744.68 °C and 829.64 °C.
• The model demonstrated superior performance compared to conventional algorithms in predicting key operational parameters.

Conclusions:

• The developed hybrid AI technique effectively predicts DLE gas turbine operating ranges, mitigating tripping issues and supporting efficient load planning.
• This predictive model serves as a valuable preventive maintenance strategy for systems requiring tight operational control.
• The findings contribute significantly to improving control strategies and ensuring the reliable operation of DLE gas turbines in the power generation sector.