지도학습을 이용한 전해질 박막 코팅 두께 형상 분류 평가 (Classification Evaluation of Electrolyte Thin Film Coating Thickness using Supervised Learning)

서지정보

· 발행기관 : 한국경영공학회
· 수록지 정보 : 한국경영공학회지 / 25권 / 4호 / 55 ~ 61페이지
· 저자명 : 김성돈, 전영호, 박종진, 배준성

초록

Purpose This paper was studied to confirm the possibility of classifying the thickness shape of the electrolyte thin film coating in the width direction using several supervised learning methods.
Methods In order to evaluate the coating uniformity, data on the shape of the coating thickness was collected through online non-destructive thickness measurement. After defining classes for coating patterns in the width direction through exploratory analysis, supervised learning algorithms such as Random Forest, SVM, KNN, GNB, and MLP were used to evaluate and compare the accuracy of the models.
Results In the case of Random Forest, SVM, KNN, and GNB algorithms, classification into 5 classes has higher accuracy than those classified into 9 classes. The MLP algorithm showed higher classification accuracy in 9 classes than in 5 classes. It was confirmed that the classification accuracy can be achieved more than 80% by adjusting the classification classConclusion The results illustrate that it is possible to classify the shape of the electrolyte thin film coating thickness in the width direction through supervised learning.

영어초록

Purpose This paper was studied to confirm the possibility of classifying the thickness shape of the electrolyte thin film coating in the width direction using several supervised learning methods.
Methods In order to evaluate the coating uniformity, data on the shape of the coating thickness was collected through online non-destructive thickness measurement. After defining classes for coating patterns in the width direction through exploratory analysis, supervised learning algorithms such as Random Forest, SVM, KNN, GNB, and MLP were used to evaluate and compare the accuracy of the models.
Results In the case of Random Forest, SVM, KNN, and GNB algorithms, classification into 5 classes has higher accuracy than those classified into 9 classes. The MLP algorithm showed higher classification accuracy in 9 classes than in 5 classes. It was confirmed that the classification accuracy can be achieved more than 80% by adjusting the classification classConclusion The results illustrate that it is possible to classify the shape of the electrolyte thin film coating thickness in the width direction through supervised learning.

참고자료

· 없음