Emission reduction in gas turbine: improving CO and NOx emission prediction using modified CNN-Bi-LSTM extrinsic attention regressor

Key Points

• Improved prediction accuracy for CO and NOx emissions occurs with a modified CNN-Bi-LSTM model.
• The regression model reduced prediction errors by over 20% across various operating conditions.
• Analyses utilized a modified convolutional neural network and bidirectional long short-term memory approach.
• This work highlights the potential of advanced neural network architectures for emission forecasting.