Machine Learning-Driven Efficiency Estimation and Variable Analysis in Combined Cycle Power Plants

Abstract

The combined cycle power plant (CCPP) has seen significant growth as a key player in the energy sector due to its efficient electricity generation and low greenhouse gas emissions. The growing global demand for electricity, fueled by rapid technological advancements, underscores the need for a reliable power supply. However, accurately predicting the efficiency of CCPPs is essential for optimizing performance and cost-effectiveness. The efficiency of power plants is influenced by a variety of environmental and internal factors, but traditional models often fail to capture these complexities. This study addresses these gaps by employing machine learning models to estimate the efficiency of a CCPP in Thailand, using a comprehensive dataset of fourteen input variables. Nine machine learning models, including regression and ensemble methods, were used for evaluation, with Random Forest Regression and Gradient Boosting achieving superior accuracy levels of 99.91% and 99.83%, respectively. Furthermore, the research delves into 14 distinct variables utilized for prediction and aims to determine which variables are of paramount significance in the assessment process.