Sparse Discovery of Nonlinear Dynamical Equations for Gas Turbine CO and NOx Emissions Using Operational Sensor Data
Faculty Mentor
Indika Ilandari Dewage
Presentation Type
Poster
Start Date
4-14-2026 9:00 AM
End Date
4-14-2026 11:00 AM
Location
PUB NCR
Primary Discipline of Presentation
Mathematics
Abstract
Accurate modeling of gas turbine emissions is essential for improving combustion efficiency and reducing environmental impact. This study applies the Sparse Identification of Nonlinear Dynamical Systems (SINDy) framework to identify governing relationships for carbon monoxide (CO) and nitrogen oxides (NOx) emissions using operational gas turbine data. The dataset includes key variables such as ambient temperature (AT), ambient pressure (AP), ambient humidity (AH), air filter differential pressure (AFDP), gas turbine exhaust pressure (GTEP), turbine inlet temperature (TIT), turbine after temperature (TAT), turbine energy yield (TEY), and compressor discharge pressure (CDP). A nonlinear library consisting of polynomial and interaction terms was constructed, and sparse regression with LASSO regularization was used to identify interpretable equations describing the dynamics of emission changes. The model was trained using data from 2011–2013 and evaluated on an independent test dataset from 2014–2015. The results show strong predictive performance, with CO test RMSE = 0.0071, MAE = 0.0049, and R² = 0.9797, and NOx test RMSE = 0.0142, MAE = 0.0107, and R² = 0.9847. The novelty of this study lies in applying a sparse dynamical systems framework to discover explicit nonlinear governing equations for both CO and NOx emissions from turbine operating variables, providing an interpretable alternative to conventional black-box machine learning models. The proposed approach offers both accurate prediction and improved physical insight into emission behavior in gas turbine systems.
Recommended Citation
Ahortu, Paul and Ilandari Dewage, indika, "Sparse Discovery of Nonlinear Dynamical Equations for Gas Turbine CO and NOx Emissions Using Operational Sensor Data" (2026). 2026 Symposium. 34.
https://dc.ewu.edu/srcw_2026/ps_2026/p1_2026/34
Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-No Derivative Works 4.0 International License.
Sparse Discovery of Nonlinear Dynamical Equations for Gas Turbine CO and NOx Emissions Using Operational Sensor Data
PUB NCR
Accurate modeling of gas turbine emissions is essential for improving combustion efficiency and reducing environmental impact. This study applies the Sparse Identification of Nonlinear Dynamical Systems (SINDy) framework to identify governing relationships for carbon monoxide (CO) and nitrogen oxides (NOx) emissions using operational gas turbine data. The dataset includes key variables such as ambient temperature (AT), ambient pressure (AP), ambient humidity (AH), air filter differential pressure (AFDP), gas turbine exhaust pressure (GTEP), turbine inlet temperature (TIT), turbine after temperature (TAT), turbine energy yield (TEY), and compressor discharge pressure (CDP). A nonlinear library consisting of polynomial and interaction terms was constructed, and sparse regression with LASSO regularization was used to identify interpretable equations describing the dynamics of emission changes. The model was trained using data from 2011–2013 and evaluated on an independent test dataset from 2014–2015. The results show strong predictive performance, with CO test RMSE = 0.0071, MAE = 0.0049, and R² = 0.9797, and NOx test RMSE = 0.0142, MAE = 0.0107, and R² = 0.9847. The novelty of this study lies in applying a sparse dynamical systems framework to discover explicit nonlinear governing equations for both CO and NOx emissions from turbine operating variables, providing an interpretable alternative to conventional black-box machine learning models. The proposed approach offers both accurate prediction and improved physical insight into emission behavior in gas turbine systems.
