Rizqi Fitri Naryanto, Mera Kartika Delimayanti, Rizky Adi, Abdurrahman, Putri Khoirin Nashiroh, Imam Sukoco, Fiqri Fadillah Fahmi, Damai Yudha Akbar Effendi, Afrilza Daffa Naryapramono
This study explores the application of machine learning regression models to predict power output in External Combustion Engine on Combined Cycle Power Plants (CCPPs) using a comprehensive dataset of 9,568 hourly observations from 2006 to 2011. Key ambient variables include temperature, pressure, humidity, and vacuum. To prevent overfitting, a 5x2 fold cross-validation strategy is employed, generating 10 unique training and testing sets. Several models are assessed, including Random Forest, XGB Regressor, Extra Trees, Hist Gradient Boosting, and LGBM Regressor. XGB Regressor demonstrates superior performance with a Mean Absolute Error (MAE) of 2.41 and Root Mean Squared Error (RMSE) of 3.37, making it the most accurate model. Additionally, the performance of ensemble models further highlights their reliability in predicting power output. The study emphasizes the importance of advanced machine learning techniques in optimizing power predictions, balancing computational efficiency, accuracy, and interpretability for large-scale industrial applications. Boosting Regressor provides a more equitable compromise between computational efficiency and performance, rendering it well-suited for implementations on a large scale. Furthermore, despite its marginally diminished accuracy, the Random Forest Regressor offers significant insights via the feature importance analysis, thereby augmenting interpretability. This study underscores the significance of sophisticated machine learning models in enhancing the precision and effectiveness of power output forecasts in CCPPs. It stresses balancing interpretability, computational cost, and accuracy in real-world applications. © Little Lion Scientific.
Mechanical Engineering Department, Engineering Faculty, Universitas Negeri Semarang, Semarang, Indonesia; Department of Computer and Informatics Engineering, Politeknik Negeri Jakarta, Depok, Indonesia; Informatic Engineering Education Department, Engineering Faculty, Universitas Negeri Semarang, Semarang, Indonesia; Automotive Engineering Education Department, Engineering Faculty, Universitas Negeri Semarang, Semarang, Indonesia; Informatic Engineering, Mathematics and Natural Sciences Faculty, Universitas Negeri Semarang, Semarang, Indonesia
Research at a Glance
Register to unlock
Topics & SDG Alignment
Register to unlock
Collaboration
Register to unlock
Author Profile (Selected)
Register to unlock
References Overview
Register to unlock
Journal & Source
Register to unlock
Metadata & Integrity
Register to unlock