Data-Driven Analysis of Gas Turbine Operating Regimes and Emission Patterns Using K-Means Clustering

The analysis of gas turbine operational data has become increasingly important for enhancing performance and mitigating environmental impact. In this study, seven months of industrial data collected from a 250 MW gas turbine operating at a production site in Algeria are investigated to identify patterns in key operating parameters and pollutant emissions. An unsupervised machine learning approach based on K-Means clustering is employed to classify turbine operating conditions into three distinct clusters. Each cluster corresponds to a specific operating regime, characterized by different levels of CO and NOx emissions as well as variations in turbine efficiency. The results demonstrate that clustering techniques provide valuable insights into the relationship between operating conditions and emission behavior, offering a practical tool for optimizing turbine performance while reducing environmental impact.

In this study, we leverage clustering analysis to analyze seven months of turbine operational data to identify patterns in emissions and operating parameters. Using K-Means clustering, the turbine operating points were grouped into three distinct clusters, each representing a specific regime with different implications for CO and NOx emission levels and overall turbine performance. These insights contribute to improved understanding of turbine behavior and support strategies for reducing environmental impact and optimizing operational efficiency based on machine learning techniques applications.