Sistem Predictive Maintenance Gripper Dispenser Penyedia Palet Pada Robot Palletizer Dengan Metode Long Short-Term Memory (LSTM)

Yusuf, Dian Maulana (2025) Sistem Predictive Maintenance Gripper Dispenser Penyedia Palet Pada Robot Palletizer Dengan Metode Long Short-Term Memory (LSTM). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 2040211011-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (9MB) | Request a copy

Abstract

Robot palletizer Fuji di Gudang Phonska IV PT Petrokimia Gresik sering mengalami kendala teknis pada sistem dispenser penyedia palet, khususnya pada komponen gripper. Permasalahan seperti ketidaksesuaian tekanan udara, rotasi gripper yang tidak maksimal, dan piston yang macet menyebabkan palet tersangkut dan mengganggu kelancaran proses pengantongan. Saat ini, penanganan masalah masih bersifat reaktif, yaitu perbaikan dilakukan setelah kerusakan terjadi, sehingga meningkatkan risiko downtime yang tidak terduga. Penelitian ini bertujuan untuk mengembangkan sistem predictive maintenance menggunakan metode Long Short-Term Memory (LSTM) untuk memprediksi potensi waktu kerusakan pada gripper dispenser sebelum kegagalan benar-benar terjadi. Sistem ini memanfaatkan data realtime dari sensor tekanan (pressure transmitter), sensor akselerometer (ADXL335) untuk rotasi grip, dan sensor posisi piston (reed switch). Data sensor yang terkumpul diolah untuk menghitung nilai availability harian sistem, kemudian nilai tersebut dinormalisasi dan dianalisis menggunakan model LSTM untuk memprediksi performa sistem hingga 30 hari ke depan. Hasil pengujian menunjukkan bahwa model LSTM dengan konfigurasi lookback window 80 hari mampu memberikan prediksi dengan tingkat akurasi yang tinggi, dibuktikan dengan nilai Mean Absolute Percentage Error (MAPE) sebesar 0,25%. Sistem akan memberikan rekomendasi jadwal perawatan jika prediksi availability turun di bawah 90%. Hasil prediksi dan rekomendasi jadwal disajikan melalui Graphical User Interface (GUI) yang informatif, sehingga memudahkan operator dalam mengambil keputusan perawatan secara proaktif dan efisien. Dengan demikian, sistem ini dapat meminimalkan downtime tidak terduga dan meningkatkan operasional di PT Petrokimia Gresik.
=====================================================================================================================================
The Fuji robot palletizer at the Phonska IV Warehouse of PT Petrokimia Gresik often experiences technical issues with its pallet dispenser system, particularly in the gripper component. Problems such as improper air pressure, suboptimal gripper rotation, and jammed pistons cause pallets to get stuck, disrupting the production process. Currently, problem-solving is reactive, with repairs performed after a failure has occurred, thereby increasing the risk of unexpected downtime. This research aims to develop a predictive maintenance system using the Long Short-Term Memory (LSTM) method to predict the potential failure time of the gripper dispenser before a breakdown actually happens. The system utilizes real-time data from a pressure transmitter, an accelerometer (ADXL335) for grip rotation, and reed switch sensors for piston position. The collected sensor data is processed to calculate the system's daily availability, this value is then normalized and analyzed using the LSTM model to predict the system's performance for up to 30 days in advance. Test results show that the LSTM model with an 80 day lookback window configuration can deliver highly accurate predictions, as evidenced by a Mean Absolute Percentage Error (MAPE) value of 0.25%. The system recommends a maintenance schedule if the predicted availability drops below 90%. The prediction results and schedule recommendations are presented through an informative Graphical User Interface (GUI), making it easier for operators to make proactive and efficient maintenance decisions. Thus, this system can minimize unexpected downtime and improve operations at PT Petrokimia Gresik.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Predictive Maintenance, Long Short Term Memory (LSTM), Gripper Dispenser, Availability, Robot Palletizer, Predictive Maintenance, Long Short-Term Memory (LSTM), Gripper Dispenser, Availability, Robot Palletizer
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
Divisions:	Faculty of Vocational > 36304-Automation Electronic Engineering
Depositing User:	Dian Maulana Yusuf
Date Deposited:	08 Aug 2025 08:01
Last Modified:	08 Aug 2025 08:01
URI:	http://repository.its.ac.id/id/eprint/128012

Actions (login required)

View Item