• 한국산학기술학회논문지
• /
• 제20권7호
• /
• pp.347-352
• /
• 2019

본 연구는 소형 유도전동기 회전자를 생산공장에서 금형주조(die casting)로 제작 시 발생하는 cast 부 품질문제를 개선하기 위하여 수행되었다. 1차 연구결과와 현장 적용성을 고려한 수많은 parametric study를 통하여 새로운 rotor core cast model을 제안하였으며, 그 제안된 model에 대한 유효성을 평가, 검증하기 위하여 주조해석 전용프로그램을 사용한 수치해석과 실제품에 대한 현장실험을 통하여 다음과 같은 결론을 얻었다. 첫째, 유도전동기 회전자의 새로운 cast model에 대한 충진해석을 수행한 결과 용탕은 하부 end ring 부를 먼저 충진하고 이동하여 상부 end ring, core slot 부분을 충진하고 최종적으로 core slot 부분에서 충진이 완료된다. 둘째, 새롭게 제안된 rotor core cast model에 대한 용탕의 유동 양상을 보면 하부 end ring 부에 용탕의 유동에 의한 와류가 적게 발생하였으며, 현장실험 결과 하부 end ring 부 단면에서 소량의 작은 결함만이 발견되었다. 셋째, 유도전동기 회전자 제작 시 rotor core cast 형상 변경을 통하여 end ring 부 결함 발생 면적이 약 70 % 감소한 양호한 품질의 회전자를 생산할 수 있다.

• 한국주조공학회지
• /
• 제42권6호
• /
• pp.369-376
• /
• 2022

다이캐스팅 공정은 다양한 산업군의 인프라 역할을 수행하는 중요한 공정이지만, 높은 불량률로 인하여 관련 기업들의 수익성 및 생산성의 한계가 있는 상황이다. 이를 타개하기 위하여, 본 연구에서는 다이캐스팅 공정의 불량 검출을 위한 산업인공지능 기반 모듈을 구성하였다. 개발된 불량 검출 모듈은 제공되는 데이터의 특징에 따라서 3단계로 동작되는 모델로 구성된다. 1단계 모델은 비지도학습 기반 이상 검출을 진행하며, 레이블이 없는 데이터셋을 대상으로 작동한다. 2단계 모델은 반지도학습 기반으로 이상 검출을 진행하며, 양품 데이터의 레이블만 존재하는 데이터셋을 대상으로 작동하며, 3단계 모델은 소수의 불량 데이터가 제공된 상황의 지도학습 모델을 기반으로 작동한다. 개발된 모델은 실제 다이캐스팅 양품 데이터를 바탕으로 96% 이상의 우수한 양품 검출 성능을 보였다.

• 한국주조공학회지
• /
• 제42권1호
• /
• pp.3-11
• /
• 2022

본 연구에서는 전산모사를 활용하여 오스테나이트계 내열강인 HR3C 합금의 수축 결함 파라미터를 확보하고, 건전부 85% 이상의 잉곳 몰드 설계 및 제작을 진행하였다. 잉곳 몰드 설계 단계에서 전산모사를 활용하여 최적의 잉곳 몰드를 설계 및 제작하였으며, 제작한 잉곳 몰드를 통해 시험조업을 수행하였다. 시험조업 후, 절단 및 비파괴 검사를 통해 결함 양상을 분석하여 수축 결함 파라미터와 잉곳 몰드 설계를 검증하였다. 검증 결과를 기반으로 HR3C 합금의 수축 결함 파라미터 (Niyama 인자, 급탕효율인자, 열간 균열 인자 등)를 확보하였으며, 확보한 파라미터를 통해 건전부 85% 이상의 잉곳 몰드 설계 방안을 확보하였다.

• 산업경영시스템학회지
• /
• 제46권4호
• /
• pp.312-320
• /
• 2023

Recently, there has been an increasing attempt to replace defect detection inspections in the manufacturing industry using deep learning techniques. However, obtaining substantial high-quality labeled data to enhance the performance of deep learning models entails economic and temporal constraints. As a solution for this problem, semi-supervised learning, using a limited amount of labeled data, has been gaining traction. This study assesses the effectiveness of semi-supervised learning in the defect detection process of manufacturing using the MixMatch algorithm. The MixMatch algorithm incorporates three dominant paradigms in the semi-supervised field: Consistency regularization, Entropy minimization, and Generic regularization. The performance of semi-supervised learning based on the MixMatch algorithm was compared with that of supervised learning using defect image data from the metal casting process. For the experiments, the ratio of labeled data was adjusted to 5%, 10%, 25%, and 50% of the total data. At a labeled data ratio of 5%, semi-supervised learning achieved a classification accuracy of 90.19%, outperforming supervised learning by approximately 22%p. At a 10% ratio, it surpassed supervised learning by around 8%p, achieving a 92.89% accuracy. These results demonstrate that semi-supervised learning can achieve significant outcomes even with a very limited amount of labeled data, suggesting its invaluable application in real-world research and industrial settings where labeled data is limited.

• Journal of Korean Neurosurgical Society
• /
• 제52권6호
• /
• pp.541-546
• /
• 2012

Objective : The prefabrication of customized cranioplastic implants has been introduced to overcome the difficulties of intra-operative implant molding. The authors present a new technique, which consists of the prefabrication of implant molds using three-dimensional (3D) printers and polymethyl-methacrylate (PMMA) casting. Methods : A total of 16 patients with large skull defects (>100 $cm^2$) underwent cranioplasty between November 2009 and April 2011. For unilateral cranial defects, 3D images of the skull were obtained from preoperative axial 1-mm spiral computed tomography (CT) scans. The image of the implant was generated by a digital subtraction mirror-imaging process using the normal side of the cranium as a model. For bilateral cranial defects, precraniectomy routine spiral CT scan data were merged with postcraniectomy 3D CT images following a smoothing process. Prefabrication of the mold was performed by the 3D printer. Intraoperatively, the PMMA implant was created with the prefabricated mold, and fit into the cranial defect. Results : The median operation time was $184.36{\pm}26.07$ minutes. Postoperative CT scans showed excellent restoration of the symmetrical contours and curvature of the cranium in all cases. The median follow-up period was 23 months (range, 14-28 months). Postoperative infection was developed in one case (6.2%) who had an open wound defect previously. Conclusion : Customized cranioplasty PMMA implants using 3D printer may be a useful technique for the reconstruction of various cranial defects.

• 한국세라믹학회지
• /
• 제36권1호
• /
• pp.61-68
• /
• 1999

층상구조를 이루는 재료의 소결시 형성되는 다양한 결함 및 잔류응력을 고찰하기 위해 TZP-SUS계 및 ZT/SUS계다층재료와 porcelain/alumina 및 porcelain/Y-TZP 이층재료를 소결법으로 제조하였다. 상압소결로 제조한 다층재료에서는 층간의 소결수축율 차이에 의해 warping, splitting, 균열 등의 소결결함이 관찰되었으며, 중간층수 및 두께의 조절과 출발물질의 제어를 통해 이러한 소결결함이 완화됨을 알 수 있었다. Tape casting법으로 제조한 다층재료에서는 소결시 가한 압력에 의해 소결결함, 특히 warping이 제어됨을 확인할 수 있었다. 이층재료에서 형성되는 잔류응력은 vickers 압입법으로 관찰하였다. Porcelain/alumina에서는 porcelain 측의 계면에 작은 인장응력이, porcelain/Y-TZP에서는 압축응력이 형성됨을 확인할 수 있었으며, 이러한 잔류응력은 이층재료의 강도에도 영향을 미침을 알수 있었다. 결국 다층재료의 소결결함 및 잔류응력은 재료설계와 출발물질 상수에 영향을 받는다는 것을 알 수 있었다.

• 한국소성가공학회:학술대회논문집
• /
• 한국소성가공학회 2001년도 추계학술대회 논문집
• /
• pp.305-308
• /
• 2001

Powder injection molding (PIM) uses the shaping advantage of injection molding but is applicable to metals and ceramics. This process combines a small quantity of polymer with an inorganic powder to form a feedstock that can be molded. After shaping, the polymeric binder is extracted and the powder is sintered often to near-theoretical densities. Accordingly, PIM delivers structural materials in a shaping technology previously restricted to polymers. The process overcomes the shape limitations of traditional powder compaction, the costs of machining, the productivity limits of isostatic pressing and slip casting, and the defect and tolerance limitations of casting. The 17-4 PH stainless steel powders with average diameter of $10{\mu}m$ were injection-molded into flat tensile specimens. Sintering of the compacts was carried out at the various temperatures ranging from 900 to $1350^{\circ}C$. Sintering behavior of the compacts and tensile properties of sintered specimens were investigated.

• 한국주조공학회지
• /
• 제24권2호
• /
• pp.94-100
• /
• 2004

Gas porosity which is a common defect in aluminum alloy casting, is also thought to be severer in aluminum alloy castings produced by lost foam process due to the pyrolysis of the polystyrene foam pattern during pouring. Fundamental experiments were carried out to evaluate the effect of process variables such as the melt treatment, the cooling rate and pouring temperature on the density and mechanical properties in A356.2 castings with simple bar shape. The density of grain refined specimen was slightly lower than that of degassed one, but was higher than that of no treated one and that of shot ball packed specimen was higher than the other specimens. The tensile strength and elongation were in the ranges of $200{\sim}230MPa$ and $0.5{\sim}1.5%$ respectively. The density and hardness of lost foam cast specimens decreased with increase in pouring temperature.

• 한국주조공학회지
• /
• 제22권1호
• /
• pp.17-25
• /
• 2002

Recently, many studies have been carried out to process on the purpose of lightness in a transport parts because of the saving energy, the environmental problem. The cast-forging process can be expected to lower costs without decreasing the mechanical properties. So, the finest microstructure is needed to get for applying the cast-forging process with Al-Si alloy because the microstructure affects to the cast-forging process. For refinement treatment of eutectic Si and Al solid-solution phase, Sr and TiB were added in Al-Si alloys. The finest microstructure could be observed when 0.075 wt.%Sr and 0.1 wt.%TiB were added respectively. In this case, tensile strength and elongation much more increased than as casting. After high temperature deformation simulation test with grain refinement specimens was carried out, about 70N per unit $area(mm^2)$ of specimen was confirmed. After hot forging, tensile strength and elongation were increased. It was considered because casting defect was removed by compressive working.

• 한국주조공학회지
• /
• 제19권1호
• /
• pp.47-53
• /
• 1999

Squeeze casting process has been used in the field of a commercial manufacturing method, in which metal is enforcedly solidified under pressure enough to prevent the cast defects such as either gas porosity or shrinkage defect. In this paper, to clarify the relationship between applied pressures and macro ${\cdot}$ microstructural behaviors in gravity and direct squeeze casts, specimens were cast by various squeezing pressures during solidification of 7000 series Al wrought alloy in the metal die designed specially. The applied pressures used in this study were 0, 25, 50, and 75 MPa. The microstructural morphologies of squeeze cast were more fine and dense with increasing the applied pressures, because of the greater solidification rate of billet resulting from the applied pressure. A normal segregation phenomenon of an increasing in amount of eutectics towards the center of the billet was observed for squeeze casts, whereas gravity cast showed an inverse segregation phenomenon of an increasing in amount of eutectics towards the edge in the billet. This change in segregation pattern which is normal or inverse is due to a higher radial temperature gradient and reduced time in the semi solid state for squeeze casting.