Studi Predictive Maintenance Pada Main Engine Kapal Berbasis Data Real-Time Condition Monitoring

Bintang, Muhammad Naufal (2026) Studi Predictive Maintenance Pada Main Engine Kapal Berbasis Data Real-Time Condition Monitoring. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6019242006-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Keandalan main engine kapal merupakan faktor penting dalam menjamin keselamatan dan efisiensi operasional pelayaran. Berdasarkan rekapitulasi data insiden main engine periode 2023–2025 di PT. X, cooling system teridentifikasi sebagai subsistem yang paling dominan menyebabkan engine breakdown. Penelitian ini bertujuan mengembangkan model predictive maintenance berbasis machine learning menggunakan algoritma Random Forest yang terintegrasi dengan sistem berbasis web milik PT. X untuk memprediksi temperatur Exhaust Gas Outlet (EGT) sebagai indikator kondisi kesehatan main engine dengan menggunakan variable data dari sensor cooling system. Penelitian diawali dengan analisis data insiden dan identifikasi komponen kritis menggunakan metode FMEA dan interview expert. Berdasarkan hasil FMEA, dipilih parameter Jacket Cooling Water (JCW) Inlet Temperature, JCW Inlet Pressure, dan JCW Outlet Temperature sebagai variabel independen, sedangkan Exhaust Gas Outlet Temperature digunakan sebagai variabel dependen. Data IoT kemudian melalui tahapan preprocessing berupa data cleansing dan feature engineering. Selanjutnya dilakukan hyperparameter tuning menggunakan RandomizedSearchCV untuk memperoleh konfigurasi Random Forest terbaik pada masing-masing silinder, kemudian model diintegrasikan ke dalam modul Predictive Analytics pada web-platform. Hasil penelitian menunjukkan bahwa model Random Forest mampu memberikan performa prediksi yang baik pada enam silinder main engine dengan nilai koefisien determinasi (R²) berkisar antara 0,8920–0,9803, nilai Mean Absolute Error (MAE) sebesar 3,14–7,39°C, dan Root Mean Square Error (RMSE) sebesar 4,27–11,17°C. Integrasi model ke dalam sistem berbasis web tersebut memungkinkan pengguna melakukan konfigurasi dataset, pelatihan model, evaluasi performa, serta visualisasi hasil prediksi melalui antarmuka web. Penelitian ini menunjukkan bahwa penerapan algoritma Random Forest berbasis data IoT berpotensi mendukung implementasi predictive maintenance dan meningkatkan keandalan operasional main engine kapal.
============================================================================================================================================
The reliability of a ship's main engine is a crucial factor in ensuring the safety and efficiency of shipping operations. Based on the recapitulation of main engine incident data for the 2023–2025 period at PT. X, the cooling system was identified as the most dominant subsystem causing engine breakdown. This study aims to develop a machine learning-based predictive maintenance model using the Random Forest algorithm integrated with PT. X's web-based system to predict the Exhaust Gas Outlet (EGT) temperature as an indicator of the main engine's health condition using variable data from the cooling system sensor. The study began with incident data analysis and identification of critical components using the FMEA method and expert interviews. Based on the FMEA results, the Jacket Cooling Water (JCW) Inlet Temperature, JCW Inlet Pressure, and JCW Outlet Temperature parameters were selected as independent variables, while the Exhaust Gas Outlet Temperature was used as the dependent variable. The IoT data then underwent preprocessing stages in the form of data cleansing and feature engineering. Furthermore, hyperparameter tuning was performed using RandomizedSearchCV to obtain the best Random Forest configuration for each cylinder, then the model was integrated into the Predictive Analytics module on the web-platform. The results of the study indicate that the Random Forest model is able to provide good predictive performance on a six-cylinder main engine with a coefficient of determination (R²) value ranging from 0.8920–0.9803, a Mean Absolute Error (MAE) value of 3.14–7.39°C, and a Root Mean Square Error (RMSE) of 4.27–11.17°C. The integration of the model into the web-based system allows users to configure the dataset, train the model, evaluate performance, and visualize the prediction results through a web interface. This study shows that the application of the Random Forest algorithm based on IoT data has the potential to support the implementation of predictive maintenance and improve the operational reliability of ship main engines.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Internet of Things, Machine Learning, Main Engine, Predictive Maintenance¸ Web-Based
Subjects:	T Technology > TJ Mechanical engineering and machinery > TJ174 Maintenance and repair of machinery T Technology > TJ Mechanical engineering and machinery > TJ217.6 Predictive Control V Naval Science > VM Naval architecture. Shipbuilding. Marine engineering > VM731 Marine Engines
Divisions:	Faculty of Marine Technology (MARTECH) > Marine Engineering > 36101-(S2) Master Theses
Depositing User:	Muhammad Naufal Bintang
Date Deposited:	04 Aug 2026 04:18
Last Modified:	04 Aug 2026 04:18
URI:	http://repository.its.ac.id/id/eprint/142854

Actions (login required)

View Item