Yusuf, Iman Dimassetya Yanuar (2025) Prediksi Emisi dengan Machine Learning sebagai Upaya Peningkatan Efisiensi Operasional guna Menurunkan Emisi CO₂ Pembangkit Listrik Tenaga Gas dan Uap (PLTGU). Masters thesis, Institut Teknologi Sepuluh Nopember.
![]() |
Text
6014231054_Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (10MB) | Request a copy |
Abstract
Pembangkit Listrik Tenaga Gas dan Uap (PLTGU) merupakan salah satu penyumbang utama emisi karbon dioksida (CO₂) akibat proses pembakaran bahan bakar fosil. Karbon dioksida merupakan kontributor utama terhadap perubahan iklim global maupun pemanasan global, sehingga diperlukan metode yang lebih akurat dalam memantau dan memprediksi tingkat emisi. Tujuan dari penelitian ini adalah untuk mengembangkan model prediksi emisi CO₂ berbasis machine learning, yang dapat membantu meningkatkan efisiensi operasional PLTGU serta mendukung pengurangan emisi secara optimal. Model ini diharapkan dapat memberikan wawasan lebih baik dalam mengelola emisi guna memenuhi regulasi lingkungan yang berlaku dengan memanfaatkan data operasional PLTGU. Penelitian ini mencakup metode pemantauan emisi berbasis Continuous Emission Monitoring System (CEMS) dan pendekatan machine learning dalam prediksi emisi. Model yang digunakan adalah Artificial Neural Network (ANN), Extreme Gradient Boosting (XGBoost), dan Random Forest (RF), yang telah banyak diterapkan dalam berbagai studi terkait optimasi pembangkitan listrik dan pengelolaan emisi gas buang. beberapa penelitian sebelumnya menunjukkan bahwa pendekatan machine learning dapat menghasilkan nilai prediksi yang lebih akurat dan presisi dibandingkan dengan metode yang sudah umum dan konvensional seperti regresi linier. Hasil penelitian menunjukkan bahwa model XGBoost dengan hyperparameter tuning merupakan model terbaik dengan nilai R² sebesar 0,9237, MAE sebesar 0,0525, dan MSE sebesar 0,0044. Implementasi model ini berhasil menurunkan intensitas emisi secara kumulatif sebesar 4,07% Optimasi ini menghasilkan penghematan bahan bakar sebesar 698.022 MMBTU per tahun atau setara dengan penghematan biaya sebesar ± Rp 66 Milyar/tahun, serta reduksi emisi sebesar 40.738,44 ton CO₂ per tahun. Temuan ini membuktikan bahwa pendekatan machine learning tidak hanya meningkatkan akurasi prediksi emisi, tetapi juga berkontribusi nyata dalam meningkatkan efisiensi operasional dan mendukung target dekarbonisasi sektor energi.
====================================================================================================================================
Combined Cycle Power Plants (CCPP) are among the major contributors to carbon dioxide (CO₂) emissions due to the combustion of fossil fuels. Carbon dioxide is a primary contributor to global climate change and global warming, making it necessary to develop more accurate methods for monitoring and predicting emission levels. The aim of this research is to develop a machine learning-based CO₂ emission prediction model that can help improve the operational efficiency of CCPP and support optimal emission reduction. This model is expected to provide better insights for emission management in compliance with applicable environmental regulations By utilizing operational data from CCPP. This study incorporates Continuous Emission Monitoring System (CEMS) for emission monitoring and applies machine learning approaches to emission prediction. The models used in this study include Artificial Neural Network (ANN), Extreme Gradient Boosting (XGBoost), and Random Forest (RF), which have been widely implemented in various studies related to power generation optimization and exhaust gas emission management. Previous research indicates that machine learning approaches can produce more accurate and precise predictions compared to traditional methods such as linear regression. The results of this study indicate that the XGBoost model with hyperparameter tuning is the best-performing model, achieving an R² value of 0.9237, a MAE of 0.0525, and an MSE of 0.0044. The implementation of this model successfully reduced emission intensity cumulatively by 4.07%. This optimization resulted in a fuel savings of 698,022 MMBTU per year, equivalent to a cost savings of approximately IDR 66 billion per year, as well as an emission reduction of 40,738.44 tons of CO₂ per year. These findings demonstrate that the machine learning approach not only improves the accuracy of emission predictions but also makes a tangible contribution to enhancing operational efficiency and supporting the decarbonization targets of the energy sector.
Item Type: | Thesis (Masters) |
---|---|
Uncontrolled Keywords: | PLTGU, Emisi CO₂, ANN, XGBoost, Random Forest, CCPP, CO₂ Emissions |
Subjects: | T Technology > TD Environmental technology. Sanitary engineering > TD883.5 Air--Pollution |
Divisions: | Faculty of Civil, Environmental, and Geo Engineering > Environmental Engineering > 25101-(S2) Master Theses |
Depositing User: | Iman Dimassetya Yanuar Yusuf |
Date Deposited: | 22 Jul 2025 08:09 |
Last Modified: | 22 Jul 2025 08:09 |
URI: | http://repository.its.ac.id/id/eprint/120488 |
Actions (login required)
![]() |
View Item |