Wihardi, Prima Ahmad (2016) Optimasi Tebal Lapisan Recast, Kekasaran Permukaan Dan Microcrack Pada Proses Pemesinan Wire Electrical Discharge Machining (WEDM) Baja Perkakas SKD61 Dengan Menggunakan Metode Optimasi Back-Propagation Neural Network Dan Genetic Algorithm. Masters thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
2111201007-Master-Theses.pdf Download (4MB) | Preview |
Abstract
Proses pemesinan wire electrical discharge machining (WEDM) adalah
suatu proses pemotongan yang didasarkan pada erosi karena bunga api listrik
untuk menghasilkan bentuk-bentuk yang kompleks dan presisi dengan
menggunakan elektroda kawat. Material benda kerja yang dipotong biasanya
bersifat konduktif. Pengaturan parameter-parameter proses pemesinan
WEDMyang tidak tepat akan menghasilkan lapisan recast yang tebal, permukaan
benda kerja yang kasar, dan microcrack yang berjumlah banyak.Penelitian ini
dilakukan untuk menentukan pengaturan level-level dari parameter-parameter
proses WEDM yang tepat agar dapat meminimalkan kekasaran permukaan benda
kerja, tebal lapisan recast dan microcrack secara serentak.
Penelitian tentang optimasi proses WEDM dilakukan pada baja perkakas
SKD61 dengan menggunakan elektroda kawat zinc coating brass. Percobaan
dilakukan dengan memvariasikan lima parameter proses WEDM, yaitu arc on
time, on time, open voltage, off time, dan servo voltage.Rancangan percobaan
ditetepkan dengan menggunakan metode Taguchi dan berupa matriks
ortogonalL18, karena parameter arcon time memiliki dua level, danparameterparameter
lainnya masing-masing memiliki tiga level. Replikasi dilakukan
sebanyak dua kali. Hal ini disebabkan karena adanya faktor noise yang tidak
dimasukkan ke dalam rancangan percobaan.Metode peramalan back-propagation
neural network (BPNN) digunakan untuk pemodelan hubungan antara parameterparameter
proses dengan parameter-parameter respon. Metode optimasi genetic
algorithm (GA) digunakan untuk menentukan seting kombinasi parameter yang
dapat meminimalkan tebal lapisan recast, kekasaran permukaan dan
microcracksecara serentak.
Model peramalan dengan BPNN yang dikembangkan memiliki arsitektur
jaringan 5-8-8-3, yang terdiri dari 5 input layer, 2 hidden layer dengan 8 neuron
pada masing-masing hidden layer, dan 3 output layer.Fungsi aktivasinya
adalahtansig dan fungsi pelatihan adalah trainrp. Optimasi GA menghasilkan
kombinasi parameter-parameter yang dapat meminimalkan tebal lapisan recast,
kekasaran permukaan dan microcracksecara serentak pada seting parameterparameter
proses arc on time sebesar 2μs,on time sebesar 0,3 μs, open voltage
sebesar 90 volt, off timesebesar 10 μs,dan servo voltage sebesar 40 volt.
===============================================================================
Wire electrical discharge machining process machining (WEDM) is a cutting
process based on the erosion due to electric sparks to produce a complex shapes and
precision by using a wire electrode.Workpiece material being cut usually is
conductive.Setting the parameters of EDM machining processes that are not
appropriate will produce a recast layers is thick, the workpiece surface is rough, and
microcracks are amounted to much.
This research is conducted to determine levels of regulation parameters appropriate
WEDM process in order to minimize the surface roughness of the workpiece,
thickness layer of recast and microcracks simultaneously.Research on the EDM
process optimization performed on SKD 61 tool steel with a zinc coating using a
brass wire electrode.Five Experiments are performed by varying the parameters of
the EDM process. Consists of the arc on time, on time, open voltage, off-time, and
servo voltage.The case of parameters of the arc on time has two levels and other has
three levels of each. The Trial design is set by using the Taguchi method and form
of orthogonal array L18.Replication is performed twicedue to of noise factor which
is not incorporated into the experimental design.
Method of forecasting backpropagation neural network (BPNN) is used for
modeling the relationship between process parameters with the parameters of the
respond.Methods of genetic optimization algorithm (GA) is used to determine the
setting combinations of parameters that can minimize the recast layer thickness,
surface roughness and microcracks simultaneously.Forecasting models developed
by BPNN has a 5-8-8-3 network architecture which are consist of 5 input layer, two
hidden layer with 8 neurons in each hidden layer and output layer 3.Activation
function is tansig and training functions are trainrp.GA optimization produces the
combination of parameters that can minimize the recast layer thickness, surface
roughness and microcracks simultaneously on setting process parameters arc on
time of 2 μs, on time of 0.3 μs, the open voltage of 90 volts, time off for 10 μs, and
the servo voltage of 40 volts.
| Item Type: | Thesis (Masters) |
|---|---|
| Additional Information: | RTM 671.53 Wih o |
| Uncontrolled Keywords: | kekasaran permukaan; lapisan recast; microcrack; BPNN; GA; surface roughness; recast layer; microcracks |
| Subjects: | T Technology > TJ Mechanical engineering and machinery > TJ230 Machine design |
| Divisions: | Faculty of Industrial Technology > Mechanical Engineering > 21101-(S2) Master Thesis |
| Depositing User: | Anis Wulandari |
| Date Deposited: | 29 May 2017 02:55 |
| Last Modified: | 27 Dec 2018 08:42 |
| URI: | http://repository.its.ac.id/id/eprint/41392 |
Actions (login required)
 |
View Item |