Skripsi

APLIKASI METODE RANDOM FOREST CLASSIFIER PADA NILAI KEKASARAN PERMUKAAN BAJA S45C PROSES MILLING CNC

---

Penilaian

---

Artificial Intelligence (AI) is a program designed to enable machines to think like humans. The utilization of AI by industry is not limited to the telecommunications sector but extends to the manufacturing field, where machine learning is employed in milling machines to classify surface roughness classes. This is particularly done for the Ra value, which is the industry standard for surface roughness used to determine the quality of a product. This research aims to apply the random forest machine learning method to classify surface roughness values obtained through the CNC milling process on S45C steel. In this study, the Random Forest method is used to facilitate the determination of surface roughness value groups based on Ra values. The accuracy calculation is conducted by comparing the real class groups with the predicted ones to determine the correctness of the class group predictions. The research utilizes the milling process by varying several parameters such as cutting speed, feed rate, and axial depth. After the workpiece is milled, the surface roughness value is measured. The classification will be carried out using Python software, where the classification is based on surface roughness value classes that have been grouped into four categories. Classification using Python is conducted to create training and test data, aiming to visualize the comparison graph between real and predicted values and to display the classification results based on coding. xx This research also employs manual calculations in the classification of surface roughness class groups, which yield accuracy values that meet standards. However, for Vc "75", entropy and information gain need to be recalculated to establish an accurate node in explaining the decision tree. The Vc "75" node cannot determine the Ra class group classification due to insufficient data, resulting in a calculation of 0 or equal values between the Fz and Ax attributes. The lack of data in the four class groups is a result of the inherent limitation of the random forest method, which requires a dataset rich in information. The calculations show that parameters with variable Vc 75 and Fz 0.05 have two decision outcomes in Ra1 and Ra4 classes, whereas variable Vc 75 and Ax 1.25 have one decision tree in Ra1 class. When all three variables are considered, the classification decision is Ra1, thus a decision tree can be constructed as shown below. Finally, accuracy comparison is conducted by printing the prediction results to determine whether the predicted class groups match the real class groups. This comparison uses variations in the number of trees: 10, 20, 30, 80, and 100, with a data train-test split of 60% and 40% respectively. The results indicate that the best prediction accuracy is 9 out of 12 test data, yielding an accuracy rate of 83.3% with 20 trees. Keywords: random forest, classification, cnc milling, surface roughness

---

Availability

Inventory Code	Barcode	Call Number	Location	Status
2407003838	T147710	T1477102024	Central Library (Reference)	Available but not for loan - Not for Loan

Detail Information

Series Title

-

Call Number

T1477102024

Publisher

Indralaya : Prodi Teknik Mesin, Fakultas Teknik Universitas Sriwijaya., 2024

Collation

ix, 57 hlm.; ilus.; tab.; 29 cm

Language

Indonesia

ISBN/ISSN

-

Classification

670.420 7

Content Type

Text

Media Type

unmediated

Carrier Type

-

Edition

-

Subject(s)

Prodi Teknik Mesin
Random Forest Classifier

Specific Detail Info

-

Statement of Responsibility

KA

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment