Wibowo, Yohanes Tri Joko (2024) Optimasi Pemesinan dengan Prioritas dan Target melalui Integrasi Jaringan Syaraf Tiruan dan Response Surface Methodology pada Pemesinan CNC Milling. Doctoral thesis, Institut Sepuluh Nopember.
Text
7010202003 -Dissertation.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (5MB) | Request a copy |
Abstract
Proses pemesinan dikontrol dengan nilai variabel pemesinan agar hasil sesuai spesifikasi. Penentuan nilai variabel pemesinan secara sembarangan mengakibatkan hasil pemesinan tidak sesuai spesifikasi sehingga produk harus diproses ulang atau dibuat kembali serta menimbulkan kerugian biaya, waktu, energi, dan kepercayaan pelanggan. Pemahaman fungsi, dampak, dan interaksi antar variabel pemesinan terhadap spesifikasi sangatlah kritis. Karena itu, pengaturan beberapa variabel pemesinan beserta tingkat prioritasnya secara benar diperlukan untuk menghindari kerugian. Prioritas memiliki pemahaman spesifikasi apa yang lebih diutamakan sedangkan target memiliki arti nilai minimal yang diperbolehkan. Pertanyaan penelitian meliputi identifikasi variabel pemesinan dan kontribusinya, observasi dampak variabel pemesinan, dan kebutuhan model optimasi multi tujuan yang menghasilkan produk sesuai spesifikasi. Mesin Computer Numerical Controll (CNC) Milling dengan kontrol Fanuc dijadikan obyek penelitian karena faktor besarnya populasi. Material PX5 digunakan karena merupakan seri perbaikan dari PX20 yang digunakan untuk produk cetakan. Variabel pemesinan hasil focused group discussion adalah kecepatan potong, laju pemakanan, kedalaman pemakanan, dan lebar pergeseran pemakanan. Tujuan penelitian untuk menentukan spesifikasi teknis dan bobotnya dicapai melalui focused group discussion dan House of Quality yang melibatkan 14 pakar pemesinan, pembuatan alat potong, injeksi plastik, dan akademisi. Bobot tujuan penelitian untuk keausan alat potong 42%, penyimpangan ukuran 38%, dan kekasaran permukaan 20%. Pada penelitian ini, metode regresi dimasukkan ke dalam kerangka kerja jaringan syaraf tiruan. Variabel pemesinan sebagai input diregresikan untuk menghasilkan respon, yang merupakan neuron pada lapisan tersembunyi. Neuron pada lapisan tersembunyi merupakan input pada regresi berikutnya sehingga menghasilkan respon yang merupakan neuron pada lapisan output. Dari simulasi numerik, regresi berbasis response surface methodology (RSM) menunjukkan kinerja terbaik berdasarkan nilai koefisien korelasi, RMSE, MAPE, dan koefisien determinasi. Model optimasi multi tujuan yang mempertimbangkan prioritas dan target dibuat menggunakan metode regresi berbasis RSM. Nilai variabel pemesinan hasil optimasi adalah kecepatan potong 71.77m/menit, laju pemakanan 5.05 mm/detik, kedalaman pemakanan 0.2mm, dan lebar pergeseran pemakanan 0.4mm. Keausan alat potong 1.62 um, penyimpangan ukuran 14.97 um, dan kekasaran permukaan 14.98 um dihasilkan dari nilai tersebut. Persentase error tertinggi pada pengujian variabel pemesinan di mesin adalah 3.05%. Dengan demikian, model optimasi yang diusulkan memberikan hasil sesuai dengan target dan dalam batas toleransi.
======================================================================================================================================
The machining process is controlled by the machining variable values to meet specifications. Determining the variable values inappropriately causes the results specifications not met and the product must be reprocessed or remade. The costs, time, energy and customer trust are lost. Understanding the function, impact and interaction between machining variables on specifications is critical. Therefore, setting the machining variables and the priority appropriately is necessary to avoid losses. Priority means understanding what specifications are more important, while targeted means the maximum value that is allowed. Research questions include the identification of machining variables and their contributions, observation of the impact of machining variables, and the need for multi-objective optimization models that produce products according to specifications. Because of the large population, the Computer Numerical Controll (CNC) Milling machine with Fanuc control was used as the research object. PX5 material is used because it is an improved PX20 series used for moulded products. The machining variables resulting from focus group discussions are cutting speed, feed rate, feed depth, and feed width shift. The research aimed to determine the technical specifications and weights achieved through focused group discussion and the house of quality, which involved 14 experts in machining, cutting tool manufacturing, plastic injection, and academics. The research objective weights for cutting tool wear were 42%, deviation 38%, and surface roughness 20%. In this research, the regression method is incorporated into an artificial neural network framework. The machining variables as input are regressed to produce the responses, which are neurons in the hidden layer. Neurons in the hidden layer as input to the next regression are regressed to produce responses, which are neurons in the output layer. From numerical simulations, regression based on response surface methodology shows the best performance based on the correlation coefficient, RMSE, MAPE and coefficient of determination values. A multi-objective optimization model that considers priorities and targets is created using an RSM-based regression method. The machining variable values resulting from the optimization are cutting speed 71.77m/minute, feed rate 5.05 mm/second, depth of cut 0.2mm, and width of cut 0.4mm. Cutting tool wear of 1.62 um, dimension deviation of 14.97 um, and surface roughness of 14.98 um resulted from these values. The highest error percentage in testing machining variables on a machine is 3.05%. Thus, the proposed optimization model provides results according to the target and within tolerance limits.
Item Type: | Thesis (Doctoral) |
---|---|
Uncontrolled Keywords: | variabel pemesinan, model optimasi multi tujuan, optimasi tertarget, optimasi prioritas, response surface methodology, machining variables, multi objective optimization model, targeted optimisation, prioritized optimization |
Subjects: | T Technology > TJ Mechanical engineering and machinery > TJ1225 Milling machines numerically controlled |
Divisions: | Faculty of Industrial Technology and Systems Engineering (INDSYS) > Industrial Engineering > 26001-(S3) PhD Thesis |
Depositing User: | Yohanes Tri Joko Wibowo |
Date Deposited: | 08 Aug 2024 12:01 |
Last Modified: | 03 Sep 2024 07:28 |
URI: | http://repository.its.ac.id/id/eprint/115081 |
Actions (login required)
View Item |