Sutawanir Darwis
Statistika, Universitas Islam Bandung

Published : 5 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 5 Documents
Search

Penerapan Metode K-Nearest Neighbors (kNN) pada Bearing Anggi Priliani Yulianto; Sutawanir Darwis
Jurnal Riset Statistika Volume 1, No. 1, Juli 2021, Jurnal Riset Statistika (JRS)
Publisher : UPT Publikasi Ilmiah Unisba

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (982.36 KB) | DOI: 10.29313/jrs.v1i1.16

Abstract

Abstract. Monitoring the condition of the engine is a top priority to avoid damage. To know the condition of the bearing, it is important to know the remaining useful life of the machine. In the IEEE PHM 2012 Prognostic Challenge platform provides real data related to accelerated bearing degradation carried out under constant operating conditions and online controlled variables of temperature and vibration (with horizontal and vertical accelerometers). In this platform, the data used is bearing2_3 data in the horizontal direction which has a duration of about 2 hours, calculated RMS every 1/10 second (2560 data). In this study machine learning based modeling will be done using the k-nearest neighbor (kNN) method to determine the prediction of RMS bearings. The kNN method is based on the classification of objects based on training data that is the closest distance to the object. kNN is a nonparametric machine learning algorithm which is a model that does not assume distribution. The advantage is that the class decision line produced by the model can be very flexible and very nonlinear. The smallest MSE value was obtained at k = 16 with MSE value = 0.157579. After getting the optimum k value, proceed with predicting a RMS of 97 lags and identifying bearing performance in several phases. Abstrak. Pemantauan kondisi mesin menjadi prioritas utama untuk menghindari adanya kerusakan. Untuk mengetahui kondisi bantalan, penting untuk mengetahui sisa masa manfaat dari mesin tersebut. Dalam platfrom IEEE PHM 2012 Prognostic Challenge ini menyediakan data nyata terkait dengan degradasi bantalan yang dipercepat yang dilakukan di bawah kondisi operasi konstan dan variabel yang dikendalikan secara online berupa suhu dan getaran (dengan akselerometer horizontal dan vertikal). Dalam platform ini, data yang digunakan adalah data bearing2_3 pada arah horizontal yang berdurasi sekitar 2 jam ini dihitung RMS setiap 1/10 detik (2560 data). Dalam penelitian ini akan dilakukan pemodelan berbasis machine learning menggunakan metode k-nearest neighbor (kNN) untuk mengetahui prediksi RMS bearing. Metode kNN didasarkan pada klasifikasi terhadap objek berdasarkan data pelatihan yang jaraknya paling dekat dengan objek tersebut. kNN merupakan salah satu algoritma pembelajaran mesin yang bersifat nonparametrik yakni model yang tidak mengasumsikan distribusi. Kelebihannya adalah garis keputusan kelas yang dihasilkan model tersebut bisa jadi sangat fleksibel dan sangat nonlinier. Nilai MSE terkecil diperoleh pada k = 16 dengan nilai MSE = 0,157579. Setelah mendapatkan nilai k optimum, dilanjutkan dengan memprediksi RMS sebanyak 97-lag serta mengidentifikasi performance kinerja bearing dalam beberapa fase.
Prediksi Sisa Umur Bearing Menggunakan Distribusi Weibull Uun Unaijah; Sutawanir Darwis
Jurnal Riset Statistika Volume 2, No. 1, Juli 2022, Jurnal Riset Statistika (JRS)
Publisher : UPT Publikasi Ilmiah Unisba

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (215.497 KB) | DOI: 10.29313/jrs.vi.909

Abstract

Abstract. The condition of the machine to avoid damage, the machine must always be monitored so that there is no decrease in operating time or unexpected damage to the machine. The condition of the health of the machine can detect, classify and predict future failures, it is very important in reducing operating and maintenance costs. There are several methods to analyze the life of the machine, one of which is the analysis using the Weibull distribution which can be used to estimate reliability, maintenance, and can be used to estimate damage. The data used in this study is secondary data obtained from the Intelligent Maintenance System (IMS), IEEE PHM 2012 through FEMTO-ST Institute storage and the Zhai Journal with the title Analysis of Time-to-Failure Data with Weibull Model in Product Life Cycle Management. Determine Time to Failure by determining the maximum value in each period. The results of data analysis from research conducted on the prediction of the remaining life of the bearing machine, it is found that the Weibull distribution can be used to analyze failure data using the smallest method based on the maximum probability and probability. However, in this case the method using the least squares method is more accurate than the maximum likelihood method. Abstrak. Pemantauan kondisi mesin untuk menghindari adanya kerusakan, mesin harus selalu dipantau agar tidak terjadi penurunan waktu operasi atau kerusakan pada mesin yang tak terduga. Kondisi dari kesehatan mesin dapat mendeteksi, mengklasifikasikan dan memperkirakan kerusakan yang akan datang, hal tersebut sangat penting dalam mengurangi biaya operasi dan pemeliharaan. Terdapat beberapa metode untuk menganalisis masa pakai mesin salah satunya analisis dengan menggunakan distribusi Weibull yang dapat digunakan untuk memperkirakan tentang persoalaan reliability, mantainability dan dapat digunakan untuk memperkirakan kerusakan bearing. Data yang digunakan pada penelitian ini adalah data sekunder yang diperoleh dari Intelligent Maintenance System (IMS), IEEE PHM 2012 melalui penyimpanan FEMTO-ST Institute dan Jurnal Zhai dengan judul Analysis of Time-to-Failure Data with Weibull Model in Product Life Cycle Management. Penentuan Time to Failure yaitu dengan menentukan nilai maksimum dalam setiap periode. Berdasarkan hasil analisis data dari penelitian yang dilakukan tentang prediksi sisa umur mesin bearing maka didapatkan bahwa distribusi Weibull dapat digunakan untuk menganalisis data waktu kegagalan dengan menggunakan estimasi metode kuadrat terkecil dan maksimum likelihood. Namun dalam hal ini metode dengan menggunakan metode kuadrat terkecil lebih akurat dibandingkan metode maksimum likelihood.
Analisis Survival pada Tungsten dan Degradasi Bearing Menggunakan Regresi Weibull Gimma Sefira Alamsya; Sutawanir Darwis
Bandung Conference Series: Statistics Vol. 1 No. 1 (2021): Bandung Conference Series: Statistics
Publisher : UNISBA Press

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (247.066 KB) | DOI: 10.29313/bcss.v1i1.37

Abstract

Abstract. Many products are needed by humans in various industries that’s why the selection of materials must be in accordance with the needs because it will affect the quality of the machine. Tungsten carbide is one of the working materials used because of its strong. Bearing is one of the results of the manufacture of tungsten carbide material that acts as a support for a shaft so that it always rotates without friction. Survival analysis will be used to determine the resistance of a machine's strength. The Weibull distribution has a small functional sample size for failure accuracy. We will discuss the application of Weibull regression with the Maximum Likelihood Estimation (MLE) as parameter estimates. This study uses data from the University of Durham, by Braiden, Green and Wright (1982) from experimental results of the stress level test of tungsten carbide cobalt alloy which produces data on stress rates and failure stresses. Tests on 5 different stress rates against twelve observations and it is proven that the stress rate has a significant effect on failure stresses. Then the next data is bearing vibration data from Prognostics and Health Management, FEMTO ST-Institute which consists of 17 bearings with horizontal and vertical directions so that in total there are 34 observations. With time to failure data as the dependent variable and speed and load as independent variables. The result is that only variable speed has a significant effect on time to failure. Abstrak. Banyak produk yang dibutuhkan manusia di berbagai industri itulah mengapa pemilihan bahan harus sesuai dengan kebutuhan karena akan mempengaruhi kualitas mesin. Tungsten karbida merupakan salah satu material kerja yang digunakan karena sifatnya yang kuat. Bearing adalah salah satu hasil pembuatan material tungsten karbida yang berperan sebagai tumpuan sebuah poros agar selalu berputar tanpa adanya gesekan. Akan digunakan analisis survival untuk mengetahui ketahanan kekuatan suatu mesin. Distribusi weibull memiliki bentuk fungsional sampel yang kecil untuk keakuratan kegagalan. Akan dibahas penerapan regresi weibull dengan metode Maximum Likelihood Estimation (MLE) sebagai taksiran parameter. Penelitian ini menggunakan data sekunder University of Durham, oleh Braiden, Green and Wright (1982) hasil eksperimen uji tingkat tekanan paduan tungsten karbida kobalt alloy yang menghasilkan data stress rates dan failure stresses. Pengujian pada 5 stress rates berbeda terhadap dua belas observasi dan terbukti bahwa stress rate berpengaruh secara signifikan terhadap failure stresses. Lalu data selanjutnya adalah data vibrasi bearing dari Prognostics and Health Management, FEMTO ST-Institute yang terdiri dari 17 bearing dengan arah horizontal dan vertikal sehingga ditotalkan ada 34 observasi. Dengan data time to failure sebagai variabel dependen serta speed dan load sebagai variabel independent. Dihasilkan bahwa hanya variabel speed berpengaruh seccara signifikan terhadap time to failure.
Visualisasi Kerusakan Bearing Menggunakan Metode Independent Component Analysis (ICA) Silvya Rahmatiara Putri; Sutawanir Darwis
Bandung Conference Series: Statistics Vol. 2 No. 2 (2022): Bandung Conference Series: Statistics
Publisher : UNISBA Press

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (510.792 KB) | DOI: 10.29313/bcss.v2i2.4236

Abstract

Abstract. Vibration is a response of a mechanical system either caused by a given excitation force or changes in operating conditions as a function of time. The force that causes this vibration can be caused by several sources such as contact/impact between moving/rotating components, rotation of an unbalanced mass, misalignment and also Bearing faults which will be the topic of this research. The data used is Bearing vibration data obtained from the Prognostics Center of Excellence (PcoE) through prognostic data storage donated by the Intelligent Maintenance System (IMS), University of Cincinnati in 2003. Principal Component Analysis (PCA) method is used to see how many components resulting from. Furthermore, the selected component from the Principal Component (PC) becomes the basis for the component results from the Independent Component Analysis (ICA) which is used to visually see the distribution of data. In this thesis presents ICA and compare with Principal Component Analysis (PCA). In the visual results of the plot of the Principal Component and Independent Component Bearing damage, it can be identified that each damage produces a different form of vibration after being reduced. Abstrak. Getaran merupakan respon dari sebuah sistem mekanik baik yang diakibatkan oleh gaya eksitasi yang diberikan maupun perubahan kondisi operasi sebagai fungsi waktu. Gaya yang menyebabkan getaran ini dapat ditimbulkan oleh beberapa sumber misalnya kontak/benturan antar komponen yang bergerak/berputar, putaran dari massa yang tidak seimbang (unbalance mass), misalignment dan juga karena kerusakan bantalan (Bearing fault) yang akan menjadi topik penelitian ini. Data yang digunakan yaitu data vibrasi Bearing yang diperoleh dari Prognostics Center of Excellence (PcoE) melalui penyimpanan data prognostik yang disumbangkan oleh Intelligent Maintenance System (IMS), University of Cincinnati pada tahun 2003. Metode Analisis Komponen Utama (AKU) digunakan untuk melihat berapa komponen yang dihasilkan. Selanjutnya komponen terpilih dari Komponen Utama (KU) menjadi dasar untuk hasil komponen dari Independent Component Analysis (ICA) yang digunakan untuk melihat sebaran data dengan visual oleh plot, sehingga menghasilkan beberapa komponen. Dalam skripsi ini akan disajikan ICA dalam statistik dan bandingkan metode ini dengan Analisis Komponen Utama (AKU). Pada hasil visual plot Komponen Utama dan Independent Component kerusakan Bearing dapat diidentifikasi bahwa pada setiap kerusakan menghasilkan bentuk getaran yang berbeda-beda setelah direduksi.
Exploring Pattern Recognition for Bearing Fault Diagnosis Sutawanir Darwis; Nusar Hajarisman; Suliadi; Achmad Widodo; Rejeki Wulan Islamiyati
Statistika Vol. 22 No. 2 (2022): Statistika
Publisher : Department of Statistics, Faculty of Mathematics and Natural Sciences, Universitas Islam Bandung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29313/statistika.v22i2.1128

Abstract

Traditional bearing sensory diagnostic include touching and hearing rely on personal experience, and for more complex system are unable to meet the needs of equipment fault diagnosis. The research on bearing fault diagnosis is developing significantly. Bearings are used in rotating machinery and most machinery failures are caused by bearing failures. The fault diagnosis of bearings is an important research area. The core of bearing fault diagnosis is the pattern recognition of fault features. The key of pattern recognition is to develop a reasonable classifier. Intelligent pattern recognition has been developed such as principal components, support vector machine, neural network. In this study, a bearing fault diagnosis based on exploring pattern recognition is proposed. The key to pattern recognition is to design a significant classifier. A number of features from bearing vibration of normal and fault bearing are extracted and processed using principal components of correlation matrix. Plot of principal components shows the visualization of normal and fault bearing and the classifier is chosen subjectively. The principal components exploration will be confirmed using least squares support vector machine. The parameter of support vector machine estimated using heuristic optimization particle swarm optimization. The proposed method can be applied in the detection of faults of bearing