Sistem Prediksi Kesehatan Mesin Conveyor Bagging Bag Turner Menggunakan Long Short-Term Memory (LSTM) di Unit Pengantongan Pupuk Urea

Herbiyan, Muhammad Hisyam (2026) Sistem Prediksi Kesehatan Mesin Conveyor Bagging Bag Turner Menggunakan Long Short-Term Memory (LSTM) di Unit Pengantongan Pupuk Urea. Diploma thesis, Institut Teknologi Sepuluh Nopember.

Text 2040221057-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Sistem penggerak Conveyor Bagging Back Turner pada Unit Pengantongan Pupuk Urea PT Petrokimia Gresik merupakan komponen penting yang berperan dalam menjaga kelancaran proses pengantongan pupuk. Gangguan pada motor induksi maupun gearbox dapat menyebabkan downtime yang berdampak pada terganggunya proses produksi. Saat ini, aktivitas pemantauan kondisi mesin masih didominasi oleh inspeksi berkala dan tindakan setelah terjadi kerusakan, sehingga perubahan kondisi kesehatan mesin belum dapat dipantau secara berkelanjutan. Oleh karena itu, penelitian ini bertujuan mengembangkan sistem pemantauan dan prediksi kesehatan mesin menggunakan metode Long Short-Term Memory (LSTM) untuk memantau kondisi motor induksi dan gearbox serta memprediksi perubahan kondisi kesehatan mesin berdasarkan data historis. Sistem memanfaatkan sensor getaran dan temperatur yang berkomunikasi melalui protokol RS485 (Modbus RTU). Data yang diperoleh diproses melalui tahapan data cleaning, deteksi kondisi berdasarkan threshold sensor, perhitungan Health Indicator (Condition Availability), normalisasi data, serta pembentukan data deret waktu sebelum digunakan sebagai masukan model LSTM. Model yang telah dilatih digunakan untuk melakukan forecasting kondisi kesehatan mesin, sedangkan keseluruhan sistem diimplementasikan pada Raspberry Pi 5 sebagai perangkat Edge Computing yang mendukung proses akuisisi, pengolahan, dan penyajian informasi kondisi mesin secara lokal. Hasil penelitian menunjukkan bahwa sistem mampu melakukan pemantauan kondisi motor induksi dan gearbox secara berkelanjutan serta memprediksi tren perubahan kondisi kesehatan mesin menggunakan model LSTM. Model menghasilkan nilai Mean Absolute Error (MAE) sebesar 7,01% dan Root Mean Squared Error (RMSE) sebesar 10,12%, yang menunjukkan kemampuan model dalam merepresentasikan tren perubahan kondisi kesehatan mesin berdasarkan data historis. Selain itu, sistem menyediakan fitur pemantauan yang menampilkan informasi kondisi sensor, nilai Health Indicator, hasil forecasting, riwayat data, serta notifikasi berdasarkan nilai threshold, sehingga dapat mendukung proses pemantauan kondisi mesin dan pengambilan keputusan pemeliharaan.
======================================================================================================================================
The Conveyor Bagging Back Turner drive system at the Urea Fertilizer Bagging Unit of PT Petrokimia Gresik is a critical component that plays an essential role in ensuring the continuity of the fertilizer bagging process. Failures of the induction motor or gearbox may result in downtime, disrupting the production process. Currently, machine condition monitoring activities are still dominated by periodic inspections and corrective actions after failures occur, making it difficult to continuously observe changes in machine Health. Therefore, this study aims to develop a machine Health monitoring and Prediction system using the Long Short-Term Memory (LSTM) method to monitor the condition of the induction motor and gearbox while predicting changes in machine Health based on Historical data. The proposed system utilizes vibration and temperature sensors communicating through the RS485 (Modbus RTU) protocol. The acquired data are processed through data cleaning, condition detection based on predefined sensor thresholds, Health Indicator (Condition Availability) calculation, data normalization, and time-series generation before being used as input to the LSTM model. The trained model is employed to forecast machine Health, while the entire system is implemented on a Raspberry Pi 5 as an Edge Computing device to support local data acquisition, processing, and machine condition information delivery. The experimental results demonstrate that the proposed system is capable of continuously monitoring the condition of the induction motor and gearbox while accurately predicting trends in machine Health using the LSTM model. The model achieved a Mean Absolute Error (MAE) of 7.01% and a Root Mean Squared Error (RMSE) of 10.12%, indicating its capability to represent machine Health trends based on Historical data. Furthermore, the system provides monitoring features that display sensor condition information, Health Indicator values, forecasting results, Historical data, and threshold-based notifications, thereby supporting machine condition monitoring and maintenance decision-making.

Item Type:	Thesis (Diploma)
Uncontrolled Keywords:	Prediksi Kesehatan Mesin, Long Short-Term Memory (LSTM), Edge Computing, Forecasting, Health Indicator. Machine Health Prediction.
Subjects:	Q Science > QA Mathematics > QA76.6 Computer programming. T Technology > TA Engineering (General). Civil engineering (General) > TA169 Reliability (Engineering) T Technology > TJ Mechanical engineering and machinery > TJ174 Maintenance and repair of machinery T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7871.674 Detectors. Sensors
Divisions:	Faculty of Vocational > 36304-Automation Electronic Engineering
Depositing User:	Muhammad Hisyam Herbiyan
Date Deposited:	05 Aug 2026 02:24
Last Modified:	05 Aug 2026 02:24
URI:	http://repository.its.ac.id/id/eprint/143978

Actions (login required)

View Item