Pemodelan Machine Learning Untuk Prediksi Konsumsi Bahan Bakar Pada Kapal Tunda

Afif, Muhammad Aditya (2025) Pemodelan Machine Learning Untuk Prediksi Konsumsi Bahan Bakar Pada Kapal Tunda. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5019201052_Undergraduate-thesis.pdf - Accepted Version Restricted to Repository staff only until 5 February 2027. Download (4MB) | Request a copy

Abstract

Penelitian ini mengembangkan model estimasi konsumsi bahan bakar pada mesin diesel kapal tunda menggunakan Support Vector Regression. Data operasional kapal yang digunakan dalam penelitian ini berasal dari laporan harian operasional mesin yang mencakup RPM, beban mesin, daya mesin, torsi mesin, jam operasonal mesin, jarak tempuh kapal dan faktor iklim seperti suhu air laut dan arah mata angin. Data tersebut dikumpulkan dan dibersihkan melalui proses data preprocessing untuk menghilangkan missing data dan outlier. Berdasarkan analisa menggunakan permutation feature importance dan correlation heatmap pada phyton, fitur yang paling berpengaruh terhadap nilai prediksi adalah rpm mesin, torsi dari mesin dan daya mesin. Untuk mengoptimalkan kinerja model prediksi, dilakukan tuning hyperparameter melalui metode grid search. Hyperparameter yang dioptimalkan termasuk nilai C, gamma, jenis kernel, dan degree. Kombinasi hyperparameter terbaik yang ditemukan adalah alpha sebesar 0.2, degree sebesar 3, gamma sebesar 4.5 dan kernel 'poly'. Evaluasi model dilakukan dengan membandingkan hasil prediksi terhadap data aktual menggunakan metrik Root Mean Squared Error (RMSE) dan R-squared (R²). Hasil pemodelan menunjukkan bahwa model memiliki akurasi prediksi yang tinggi dengan nilai MAE sebesar 7.2525, nilai RMSE sebesar 12.4933 dan R² sebesar 0.9622. Selain itu, Model ini dapat diterapkan pada jenis kapal tunda lainnya dengan menyesuaikan variable yang dibutuhkan dan tersedia.
========================================================================================================================
This research develops a fuel consumption estimation model for tugboat diesel engines using Support Vector Regression. The ship operational data used in this study comes from the daily engine operational report which includes RPM, engine load, engine power, engine torque, engine operating hours, ship mileage and climatic factors such as sea water temperature and wind direction. The data was collected and cleaned through data preprocessing to remove missing data and outliers. Based on the analysis using permutation feature importance and correlation heatmap in phyton, the features that have the most influence on the prediction value are engine rpm, torque of the engine and engine power. To optimize the performance of the prediction model, hyperparameter tuning is performed through the grid search method. The optimized hyperparameters include C value, gamma, kernel type, and degree. The best hyperparameter combination found was alpha of 0.2, degree of 3, gamma of 4.5 and kernel 'poly'. Model evaluation is done by comparing the prediction results to the actual data using Root Mean Squared Error (RMSE) and R-squared (R²) metrics. The modelling results show that the model has high prediction accuracy with MAE value of 7.2525, RMSE value of 12.4933 and R² of 0.9622. In addition, this model can be applied to other types of tugboats by adjusting the required and available variables.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Kata kunci: Konsumsi Bahan Bakar, Kapal Tunda, Machine Learning, Hyperparameter Tuning, Support Vector Regression Keywords: Fuel Consumption, Tugboat, Machine Learning, Hyperparameter Tuning, Support Vector Regression
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. V Naval Science > VM Naval architecture. Shipbuilding. Marine engineering > VM464 Towboats. Tugboats V Naval Science > VM Naval architecture. Shipbuilding. Marine engineering > VM731 Marine Engines
Divisions:	Faculty of Marine Technology (MARTECH) > Marine Engineering > 36202-(S1) Undergraduate Thesis
Depositing User:	Muhammad Aditya Afif
Date Deposited:	05 Feb 2025 11:18
Last Modified:	05 Feb 2025 11:41
URI:	http://repository.its.ac.id/id/eprint/118413

Actions (login required)

View Item