Ardiansyah, Mohamad Yusuf (2026) Sistem Diagnosis Kerusakan Gear Pump Dan Selang Pada Hydraulic Oil Tank Menggunakan Metode Artificial Neural Network (ANN). Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 2040221127-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
2040221127-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (7MB) |
Abstract
Hydraulic oil tank pada mesin Polyurethane di PT Toyota Boshoku Indonesia memiliki peran penting dalam menjaga kestabilan sistem hidrolik. Namun, kerusakan pada gear pump dan selang sering menyebabkan alarm mesin dan line stop karena proses diagnosis masih dilakukan secara manual dan bergantung pada alarm tekanan. Penelitian ini bertujuan merancang dan mengimplementasikan sistem diagnosis kerusakan hydraulic oil tank menggunakan metode Artificial Neural Network (ANN) berdasarkan data multisensor berupa tekanan oli hidrolik, suhu oli, getaran motor, dan arus motor. Data sensor diproses melalui segmentasi sinyal, ekstraksi fitur domain waktu, dan normalisasi, kemudian diklasifikasikan menggunakan model ANN bertipe Multilayer Perceptron. Berdasarkan hasil pengujian variasi hyperparameter, konfigurasi terbaik diperoleh menggunakan dua hidden layer berjumlah 64 dan 32 neuron dengan fungsi aktivasi Leaky ReLU. Hasil pengujian menunjukkan bahwa model mampu mencapai akurasi pengujian sebesar 98,33% dalam mengklasifikasikan tiga kondisi, yaitu normal, selang bocor, dan gear pump retak. Sistem juga mampu menyelesaikan proses diagnosis dalam waktu 58,22 detik, sehingga mendukung identifikasi kerusakan secara cepat. Hasil penelitian menunjukkan bahwa metode ANN efektif diterapkan sebagai sistem diagnosis dini kerusakan pada hydraulic oil tank untuk membantu teknisi mempercepat identifikasi gangguan serta meminimalkan dampak terhadap proses produksi.
=====================================================================================================================================
The hydraulic oil tank in the Polyurethane machine at PT Toyota Boshoku Indonesia plays a critical role in maintaining the stability of the hydraulic system. However, failures of the gear pump and hose frequently trigger machine alarms and production line stoppages because fault diagnosis is still performed manually and relies primarily on pressure alarms. This study aims to design and implement a fault diagnosis system for the hydraulic oil tank using the Artificial Neural Network (ANN) method based on multi sensor data, including hydraulic oil pressure, oil temperature, motor vibration, and motor current. The sensor data were processed through signal segmentation, time-domain feature extraction, and normalization before being classified using a Multilayer Perceptron (MLP) based ANN model. Based on hyperparameter evaluation, the optimal configuration consisted of two hidden layers with 64 and 32 neurons using the Leaky ReLU activation function. The experimental results showed that the proposed model achieved a testing accuracy of 98.33% in classifying three operating conditions: normal, hose leakage, and cracked gear pump. Furthermore, the system completed the fault diagnosis process within 58.22 seconds, enabling rapid fault identification. These findings demonstrate that the ANN-based approach is effective for early fault diagnosis of the hydraulic oil tank, assisting maintenance technicians in accelerating fault identification while minimizing the impact of equipment failures on the production process.
Actions (login required)
 |
View Item |