멀티모달 딥러닝 기반 블로우 성형 공정 프리폼 온도 예측 모델 개발
Development of preform temperature prediction model for blow molding process based on multi-modal deep learning
1
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
This study proposes a multi-modal deep learning model to predict preform temperature distribution which is a key factor affecting quality of plastic products blown in the non-conformal temperature control Injection stretch blow molding process. The model integrates thermal images of preform, heating process parameters(lamp temperatures), and environmental factors(heater and fan ambient temperatures) using a combined architecture of 2D CNN, 1D CNN, and ANN. By applying loss function based on wMAE and Dropout layers, the model achieved a MAE of 3.7353℃, showing improved accuracy and potential for smart manufacturing systems.
Abstract
This study proposes a multi-modal deep learning model to predict preform temperature distribution which is a key factor affecting quality of plastic products blown in the non-conformal temperature control Injection stretch blow molding process. The model integrates thermal images of preform, heating process parameters(lamp temperatures), and environmental factors(heater and fan ambient temperatures) using a combined architecture of 2D CNN, 1D CNN, and ANN. By applying loss function based on wMAE and Dropout layers, the model achieved a MAE of 3.7353℃, showing improved accuracy and potential for smart manufacturing systems.