Deteksi Kerusakan Bearing Menggunakan Komponen Utama Kernel
DOI:
https://doi.org/10.29313/jrs.v3i1.1771Keywords:
Deteksi Kerusakan Bearing, Komponen Utama Kernel, Vibrasi BearingAbstract
Abstract. Various factors can cause bearing damage, this is a major problem because it can cause substansial losses and affect production schedules. It is recommended to control bearing damage in order to provide early information about the damage. The method used in this paper is Hotelling T2 through Kernel Principal Component Analysis(KPCA). The principal component of kernel used to construct Hotelling T2 statistics in order to obtain T2 statistical values for detection bearing damage. The purpose of this paper is to apply KPCA as a statistic to use normal and damaged bearing. Data from NASA Bearing Dataset that contains normal bearing and damage bearing. Data divided into 1153 data for training and 323 data for testing. In this paper, 8 features are used as input for KPCA, reduced to 6 kernel principal component. Training data can obtained eigenvalue as input Hotelling T2, testing is used to detect bearings condition. Bearings are detected to damaged when T2 > 12.6939 with a significance level of 5%, from 323 training data 294 observations detected as damaged bearings and 29 observations detected as normal bearings. After conducting research, KPCA method can use to detect bearing damage.
Abstrak. Berbagai faktor dapat menyebabkan kerusakan bearing, hal ini merupakan masalah utama karena dapat menyebabkan kerugian cukup besar dan mempengaruhi jadwal produksi. Dianjurkan mengontrol kerusakan bearing agar dapat memberikan informasi awal kerusakan bearing. Metode digunakan dalam penelitian Hotelling T2 melalui komponen utama kernel. Komponen utama kernel digunakan untuk mengkontruksi statistik Hotelling T2 sehingga diperoleh nilai statistik T2 untuk deteksi kerusakan bearing. Tujuan penelitian adalah menerapkan komponen utama kernel sebagai statistik untuk menggunakan Hotelling T2 bearing normal dan bearing rusak. Data sekunder NASA Bearing Dataset berisikan data bearing normal serta data bearing rusak. Proses deteksi bearing melalui komponen utama kernel dibagi menjadi data training sebanyak 1153 data sedangkan data testing sebanyak 323 data. Berdasarkan hasil penelitian, 8 fitur dijadikan sebagai input komponen utama kernel, direduksi menjadi enam komponen utama kernel. Menggunakan data training diperoleh nilai eigen sebagai input Hotelling T2, data testing digunakan untuk mendeteksi kondisi bearing. Bearing terdeteksi rusak ketika statistik T2 > 12.6939. Dengan taraf signifikansi 5 %, didapatkan sebanyak 294 data pengamatan dari 323 terdeteksi merupakan bearing rusak sedangkan sisanya 29 data pengamatan terdeteksi merupakan bearing normal. Setelah dilakukan penelitian metode komponen utama kernel layak digunakan untuk mendeteksi kerusakan bearing.
References
M. Fathurrohman, R. G. Lulus Lambang H, and D. Djoko Susilo, “Diagnosa Kerusakan Bantalan Bola Menggunakan Metode Support Vector Machine,” 2019.
M. Devega, “Diagnosis Kerusakan Bantalan Gelinding Menggunakan Metode Radial Basis Function Neural Network (RBFNN),” 2013.
H. Qian, Y. Liu, and P. Lv, “Kernel Principal Components Analysis for Early Identification of Gear Tooth crack,” IEEE, 2006.
B. SchölkopfSch and A. M. Smola Klaus-Robert, Communicated by Peter Dayan Nonlinear Component Analysis as a Kernel Eigenvalue Problem. 1998.
J. M. Lee, C. K. Yoo, S. W. Choi, P. A. Vanrolleghem, and I. B. Lee, “Nonlinear process monitoring using kernel principal component analysis,” Chem Eng Sci, vol. 59, no. 1, pp. 223–234, 2004, doi: 10.1016/j.ces.2003.09.012.
Y. Wang and F. Deng, “A sensor fault diagnosis method based on KPCA and contribution graph,” Vibroengineering PROCEDIA, vol. 33, pp. 6–10, Oct. 2020, doi: 10.21595/vp.2020.21620.
R. Wilkinson, “Applied Multivariate Statistics: Hotelling’s T^2 distribution.”
F. Shahzad, Z. Huang, and W. H. Memon, “Process Monitoring Using Kernel PCA and Kernel Density Estimation‐Based SSGLR Method for Nonlinear Fault Detection,” Applied Sciences (Switzerland), vol. 12, no. 6, Mar. 2022, doi: 10.3390/app12062981.
Mas’ad, H. Yasin, and D. A. I. Maruddani, “Analisis Faktor-Faktor yang Mempengaruhi Persentase Penduduk Miskin di Jawa Tengah Dengan Metode Geographically Weighted Principal Components Analysis (GWPCA) Adaptive Bandwidth,” Jurnal Gaussian, vol. 5, no. 3, pp. 487–496, 2016.
U. Unaijah and S. Darwis, “Prediksi Sisa Umur Bearing Menggunakan Distribusi Weibull,” Jurnal Riset Statistika, pp. 73–81, Jul. 2022, doi: 10.29313/jrs.vi.909.
D. A. Awitasari, “Peta Kendali Multivariat Hotelling T2 Untuk Pengendalian Kualitas Nilai Mahasiswa Pada Empat Mata Kuliah Inti Di Jurusan Statistika Universitas Islam Indonesia,” 2018.