Perancangan Sistem Deteksi Kerusakan Pada Pompa Sentrifugal di PT. Saka Energi Indonesia Ltd. Dengan Menggunakan Discriminant Feature Extraction

Husna, Asma'ul (2021) Perancangan Sistem Deteksi Kerusakan Pada Pompa Sentrifugal di PT. Saka Energi Indonesia Ltd. Dengan Menggunakan Discriminant Feature Extraction. Masters thesis, Institut Teknologi Sepuluh Nopember.

[img] Text
Laporan_Asma'ul Husna_02311950010013.pdf - Accepted Version
Restricted to Repository staff only until 1 October 2023.

Download (2MB) | Request a copy

Abstract

Proses pemantauan kondisi pompa sentrifugal erat kaitannya dengan deteksi dan diagnosis kesalahan. Biasanya digunakan sinyal getaran, terutama untuk mesin yang berputar. Namun, dalam kondisi tertentu tidak mungkin memasang akselerometer pada mesin karena kondisi dan lingkungan tertentu. Sinyal arus dapat digunakan untuk deteksi kesalahan menggantikan sinyal getaran. Fitur statistik dari sinyal arus berisi informasi tentang karakteristik sinyal. Namun, fitur statistik dalam domain waktu tidak cukup sensitif terhadap gejala kesalahan yang lemah, sehingga mempengaruhi deteksi kesalahan dan akurasi klasifikasi. Sehingga digunakan fitur statistik dalam domain frekuensi, karena spektrum frekuensi lebih sensitif terhadap kesalahan yang lemah. Dalam penelitian ini dilakukan ekstraksi discriminant feature untuk mendeteksi kerusakan pada pompa sentrifugal. Discriminant feature berisi fitur-fitur sinyal dalam domain waktu dan frekuensi. Proses ekstraksi discriminant feature dibagi menjadi tiga fase. Pada fase pertama, sinyal kondisi sehat dikorelasikan silang dengan sinyal arus pompa sentrifugal di beberapa kelas kesalahan. Pada fase kedua, fitur statistik diekstraksi dari sinyal kondisi normal, kebocoran seal dan sindrom kavitasi dalam domain waktu dan frekuensi. Hasil fitur-fitur tersebut akan dikombinasikan menjadi pool discriminant feature. Pool discriminant feature akan digunakan sebagai input dalam pembuatan classifier untuk sistem deteksi kerusakan pada pompa sentrifugal. Dalam penelitian ini digunakan dua variabel pengukuran untuk perbandingan yakni arus dan kecepatan vibrasi bearing. Sistem deteksi kerusakan pada pompa sentrifugal dengan algoritma kNN menggunakan data kecepatan vibrasi bearing memiliki performansi lebih tinggi dibandingkan dengan menggunakan data arus yaitu akurasi, presisi, recall, F1 score masing-masing sebesar 99,38%, 99,38%, 99,17%, 99,27%. Sedangkan sistem deteksi menggunakan data kecepatan vibrasi bearing memiliki nilai akurasi, presisi, recall, F1 score masing-masing sebesar 99,59%, 99,59%, 99,45%, 99,44%. ====================================================================================================== The process of monitoring the condition of a centrifugal pump is closely related to fault detection and diagnosis. Vibration signals are commonly used, especially for rotating machines. However, under certain conditions, it is not possible to install an accelerometer on the engine due to certain conditions and environments. The current signal can be used for fault detection in place of vibration signals. The statistical features of the current signal contain information about the characteristics of the signal. However, statistical features in the time domain are not sensitive enough to weak error symptoms, thus affecting error detection and classification accuracy. So that statistical features in the frequency domain are used because the frequency spectrum is more sensitive to weak errors. In this study, the discriminant feature extraction has carried out to detect damage to the centrifugal pump. The discriminant feature contains signal features in time and frequency domains. The discriminant feature extraction process has divided into three phases. In the first phase, the healthy condition signal is cross-correlated with the centrifugal pump current signal in several fault classes. In the second phase, the statistical features have extracted from normal condition signals, seal leakage, and cavitation syndrome in the time and frequency domains. The result of these features will be combined into a discriminant feature pool. The pool discriminant feature will be used as input in making a classifier for the centrifugal pump damage detection system. In this study, two measurement variables are used for comparison, namely current and bearing vibration speed. The damage detection system on centrifugal pumps with the kNN algorithm using vibration speed data has a higher performance than current data. The value of accuracy, precision, recall and F1-score using current data on centrifugal pump fault detection respectively are 99,38%, 99,38%, 99,17%, 99,27%. While, the accuracy, precision, recall and F1-score using motor bearing speed vibration data on centrifugal pump fault detection respectively are 99,59%, 99,59%, 99,45%, 99,44%.

Item Type: Thesis (Masters)
Uncontrolled Keywords: Deteksi kerusakan, Discriminant feature, Pompa sentrifugal, Centrifugal pump, fault detection
Subjects: T Technology > T Technology (General) > T174 Technological forecasting
T Technology > T Technology (General) > T57.62 Simulation
T Technology > TA Engineering (General). Civil engineering (General) > TA169.5 Failure analysis
T Technology > TA Engineering (General). Civil engineering (General) > TA355 Vibration.
T Technology > TJ Mechanical engineering and machinery > TJ217.6 Predictive Control
T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK2785 Electric motors, Induction.
Divisions: Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30101-(S2) Master Thesis
Depositing User: Asma'ul Husna
Date Deposited: 24 Aug 2021 08:23
Last Modified: 24 Aug 2021 08:23
URI: https://repository.its.ac.id/id/eprint/89914

Actions (login required)

View Item View Item