Efendi, Maharani Putri (2025) Prediksi Fitur Health Index Bearing Primary Air Fan pada Sistem Boiler PLTU Menggunakan Fitur Nonlinier dan Metode Gated Recurrent unit. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5026211095-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5026211095-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (7MB) | Request a copy |
Abstract
Pembangkit Listrik Tenaga Uap merupakan tulang punggung pasokan energi di Indonesia, memberikan kontribusi hingga 50% dari total kapasitas pembangkit listrik nasional. Bearing pada Primary Air Fan (PAF) memegang peran penting dalam menjaga stabilitas rotasi poros kipas pada sistem boiler. Pemantauan kondisi bearing PAF masih didominasi metode konvensional yang tidak mampu memprediksi kerusakan secara akurat sehingga meningkatkan risiko downtime dan biaya operasional. Tugas akhir ini memberikan solusi berupa metode prediktif menggunakan analisis fitur nonlinier dan algoritme Gated Recurrent Unit (GRU). Data sensor suhu, tekanan udara, dan getaran dikumpulkan dari sistem Distributed Control System (DCS) dan diolah menggunakan Detrended Fluctuation Analysis (DFA) untuk menghasilkan fitur health index yang menjadi input bagi model GRU. Proses pengujian dilakukan dengan mengatur kombinasi hyperparameter seperti jumlah unit, learning rate, dropout, dan batch size untuk menemukan konfigurasi model terbaik. Hasil percobaan menunjukkan model GRU memiliki kemampuan prediksi yang baik dengan MAPE terendah sebesar 7,99% pada Motor Side A dan 8,73% pada Free Side A. Sementara itu, Free Side B dan Motor Side B menghasilkan MAPE terendah sebesar 9,06% dan 9,19%. Tugas akhir ini diharapkan dapat meningkatkan efisiensi pemeliharaan prediktif, mengurangi downtime, serta menekan biaya operasional pada sistem boiler PLTU.
====================================================================================================================================
Steam Power Plants are the backbone of Indonesia's energy supply, contributing up to 50% of the total national power generation capacity. Primary Air Fan (PAF) bearings play a crucial role in maintaining the stability of the fan shaft rotation in the boiler system. Monitoring of PAF bearing conditions is still dominated by conventional methods that are unable to accurately predict damage, thereby increasing the risk of downtime and operational costs. This final project offers a solution in the form of a predictive method using nonlinear feature analysis and the Gated Recurrent Unit (GRU) algorithm. Temperature, air pressure, and vibration sensor data are collected from the Distributed Control System (DCS) and processed using Detrended Fluctuation Analysis (DFA) to generate health index features that serve as input for the GRU model. The testing process is carried out by adjusting a combination of hyperparameters such as the number of units, learning rate, dropout, and batch size to find the best model configuration. The experimental results show that the GRU model has good predictive capability with the lowest MAPE of 7.99% on Motor Side A and 8.73% on Free Side A. Meanwhile, Free Side B and Motor Side B produce the lowest MAPE of 9.06% and 9.19%. This final project is expected to improve predictive maintenance efficiency, reduce downtime, and reduce operational costs in the PLTU boiler system.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | health index, detrended fluctuational analysis, gated recurrent unit, pemeliharaan prediktif, pembangkit listrik tenaga uap, health index, detrended fluctuation analysis, gated recurrent unit, predictive maintenance, steam power plant |
| Subjects: | T Technology > T Technology (General) > T174 Technological forecasting |
| Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 57201-(S1) Undergraduate Thesis |
| Depositing User: | Maharani Putri Efendi |
| Date Deposited: | 28 Jul 2025 08:12 |
| Last Modified: | 28 Jul 2025 08:12 |
| URI: | http://repository.its.ac.id/id/eprint/122616 |
Actions (login required)
 |
View Item |