Studi Analisis Performa Model Machine Learning Forecasting untuk Prediksi Kegagalan Peralatan Coal Mill pada PT PLN Nusantara Power Unit Pembangkit Indramayu

Noviali, Rahmad (2025) Studi Analisis Performa Model Machine Learning Forecasting untuk Prediksi Kegagalan Peralatan Coal Mill pada PT PLN Nusantara Power Unit Pembangkit Indramayu. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6032231057-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2027. Download (8MB) | Request a copy

Abstract

PT PLN Nusantara Power terus mendorong penerapan teknologi Industri 4.0, termasuk digitalisasi dan otomatisasi sistem pembangkit listrik, untuk meningkatkan efisiensi dan keandalan operasional. Salah satu tantangan utama di Unit Pembangkit Indramayu adalah kegagalan pada Coal Mill, yang menjadi penyumbang utama kerugian output berdasarkan analisis Pareto Loss Output (PLO) tahun 2021. Kegagalan ini menyebabkan downtime signifikan dan menurunkan nilai Mean Time Between Failure (MTBF) peralatan secara keseluruhan.
Model machine learning berbasis algoritma Long Short-Term Memory (LSTM) dan Gradient Boosting Classifier diterapkan untuk memprediksi potensi plugging (penyumbatan) pada Coal Mill. Data historis dari 14 sensor utama, yang mencakup parameter seperti arus motor, suhu, dan rasio bahan bakar terhadap udara, digunakan sebagai input. Model LSTM dirancang untuk memberikan prediksi numerik terkait kondisi operasi Coal Mill serta memberikan peringatan dini. Gradient Boosting Classifier memverifikasi hasil prediksi LSTM dengan mengklasifikasikan kondisi operasional menjadi normal atau potensi plugging.
Model diuji dengan pembagian data 80% untuk pelatihan dan 20% untuk pengujian. Validasi dilakukan menggunakan metrik Root Mean Square Error (RMSE) untuk LSTM, yang menghasilkan nilai 0,88, dan F1 Score untuk Gradient Boosting Classifier, yang mencapai 0,99. Nilai R² (R Square) sebesar 0,89 menunjukkan bahwa model LSTM mampu menjelaskan 89% variabilitas data. Hasil pengujian menunjukkan bahwa model dapat memberikan peringatan dini sekitar 20 menit sebelum kegagalan terjadi, membantu meningkatkan nilai MTBF dan mengurangi risiko downtime. Implementasi model ini mengidentifikasi variabel sensor yang berkontribusi signifikan, dan memvalidasi akurasi prediksi. Model ini diharapkan dapat meningkatkan keandalan dan efisiensi operasional Coal Mill di PLTU Indramayu.
==================================================================================================================================
PT PLN Nusantara Power continues to promote the implementation of Industry 4.0 technologies, including the digitalization and automation of power plant systems, to enhance operational efficiency and reliability. One of the main challenges faced by the Indramayu Power Plant Unit is failures in the Coal Mill, which significantly contribute to output losses based on the Pareto Loss Output (PLO) analysis in 2021. These failures result in substantial downtime and reduce the Mean Time Between Failure (MTBF) of the equipment overall.
A machine learning model based on Long Short-Term Memory (LSTM) and Gradient Boosting Classifier algorithms was applied to predict potential plugging (blockage) in the Coal Mill. Historical data from 14 key sensors, capturing parameters such as motor current, temperature, and fuel-to-air ratio, were used as input. The LSTM model was designed to provide numerical predictions related to Coal Mill operational conditions and deliver early warnings. The Gradient Boosting Classifier verified LSTM predictions by classifying operational conditions as either normal or indicative of potential plugging.
The model was tested with an 80% data split for training and 20% for testing. Validation was conducted using the Root Mean Square Error (RMSE) metric for LSTM, which yielded a value of 0.88, and the F1 Score for Gradient Boosting Classifier, which reached 0.99. An R² (R Square) value of 0.89 indicated that the LSTM model could explain 89% of data variability. Test results showed that the model could provide early warnings approximately 20 minutes before a failure occurred, enhancing MTBF values and reducing downtime risks. The model implementation identified key sensor variables contributing significantly, validated prediction accuracy, and is expected to improve the reliability and operational efficiency of the Coal Mill at the Indramayu Power Plant.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Industrial 4.0, Digitalisasi, Machine Learning, Coal Mill, LSTM, Gradient Boosting Classifier, Keandalan Opersional, MTBF, Industrial 4.0, Digitalization, Machine Learning,LSTM, Gradient Boosting Classifier , Coal Mill, Operational Reliability, MTBF
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines.
Divisions:	Interdisciplinary School of Management and Technology (SIMT) > 61101-Master of Technology Management (MMT)
Depositing User:	Rahmad Noviali
Date Deposited:	02 Feb 2025 03:27
Last Modified:	02 Feb 2025 03:28
URI:	http://repository.its.ac.id/id/eprint/117546

Actions (login required)

View Item