• 응용통계연구
• /
• 제34권1호
• /
• pp.99-114
• /
• 2021

반도체 제조 산업에서는 Big Data에 기초한 Smart Factory 도입과 적용이 가시화되면서 생산 공정의 각 단계에서 수집 가능한 다양한 센서(sensor) 데이터를 활용하여 공정 이상 탐지 및 최종 수율 예측 등에 다양한 분석 방법을 시도하고 있다. 현재 반도체 공정은 원료인 잉곳(ingot)에서 패키징(packaging) 작업 이전의 웨이퍼(wafer) 생산까지 500 600개 이상의 세부 공정과 이와 연계된 수천 개의 계측 공정으로 구성된다. 개별 계측 공정 내의 실제 계측 비율은 대상 제품 대비 0.1%에서 최대 5%를 넘지 못하고 계측 시점별로 일정하게 유지할 수 없다. 이러한 이유로 공정 각 단계의 정상 상태를 간접적으로 판단할 수 있는 장비 센서(sensor) 데이터를 활용하여 관리 여부를 판단하고자 하는 노력이 계속되고 있다. 본 연구에서는 장비 센서 데이터 기반의 공정 이상 탐지 프로세스를 정의하고 현재 적용 되고 있는 기술 통계량 기반 진단 방법의 단점을 보완하기 위해 FDA(Functional Data Analysis)방법을 활용하였다. 실제 현장 사례 데이터에 머신러닝을 이용하여 이상 탐지 정확도 비교를 통해 효과성을 검증하였다.

• 한국기계가공학회지
• /
• 제19권7호
• /
• pp.7-15
• /
• 2020

Recently, the fourth industrial revolution has been presented as a new paradigm and additive manufacturing (AM) has become one of the most important topics. For this reason, process monitoring for each cross-sectional layer of additive metal manufacturing is important. Particularly, deep learning can train a machine to analyze, optimize, and repair defects. In this paper, image classification is proposed by learning images of defects in the metal cross sections using the convolution neural network (CNN) image labeling algorithm. Defects were classified into three categories: crack, porosity, and hole. To overcome a lack-of-data problem, the amount of learning data was augmented using a data augmentation algorithm. This augmentation algorithm can transform an image to 180 images, increasing the learning accuracy. The number of training and validation images was 25,920 (80 %) and 6,480 (20 %), respectively. An optimized case with a combination of fully connected layers, an optimizer, and a loss function, showed that the model accuracy was 99.7 % and had a success rate of 97.8 % for 180 test images. In conclusion, image labeling was successfully performed and it is expected to be applied to automated AM process inspection and repair systems in the future.

• 대한기계학회:학술대회논문집
• /
• 대한기계학회 2001년도 춘계학술대회논문집C
• /
• pp.140-145
• /
• 2001

Lasting machines for shoe manufacturing are continuously developed with the aid of automation and Computer Aided Manufacturing (CAM). Adaptive lasting machine and CAD data of a shoe last are inevitably introduced for the labor-free manufacturing process. Recently, method for the CAD datarization of a shoe last is suggested using finite element mesh system. Initial set up data and control data of machine parts are required for the adaptive lasting machine. For the efficient process, grading of those data is essential to minimize data storage and production costs. In this paper, bonding lines are extracted from the CAD data of a shoe last and graded by the geometric grading system. Tension spline method is adopted for the interpolation of last CAD data. The results are compared with the results from the arithmetic grading system that is widely adopted in the shoemaking companies.

• International Journal of Internet, Broadcasting and Communication
• /
• 제12권4호
• /
• pp.109-121
• /
• 2020

As a decrease in labor became a serious issue in the manufacturing industry, smart factory technology, which combines IT and the manufacturing business, began to attract attention as a solution. In this study, we have designed and implemented a real-time remote management system for smart factories, which is connected to an IoT sensor and gateway, for plastic manufacturing plants. By implementing the REST API in which an IoT sensor and smart gateway can communicate, the system enabled the data measured from the IoT sensor and equipment status data to the real-time monitoring system through the gateway. Also, a web-based management dashboard enabled remote monitoring and control of the equipment and raw material processing status. A comparative analysis experiment was conducted on the suggested system for the difference in processing speed based on equipment and measurement data number change. The experiment confirmed that saving equipment measurement data using cache mechanisim offered faster processing speed. Through the result our works can provide the basic framework to factory which need implement remote management system.

• 한국공작기계학회:학술대회논문집
• /
• 한국공작기계학회 2001년도 춘계학술대회 논문집(한국공작기계학회)
• /
• pp.387-392
• /
• 2001

This paper is to model a 30-shape product applying mathematically the data acquired from a 3D scanner and using an Automatic Design Program. The research studied in th reverse engineering up to now has been developed continuously and surprisingly. However, forming 3D-shape solid models in CAE and CAM, based on the research, the study leaves much to be desired. Especially, analyses and studies reverse-designing automatically using measured data after manufacturing. Consequently, we are going to acquire geometric data using an 3D scanner in this study with which we will open a new field of reverse engineering by a program which can design a 3D-shape solid model in a CAD-based program automatically.

• 한국정밀공학회지
• /
• 제13권2호
• /
• pp.37-47
• /
• 1996

This paper presents an automatic tele-inspection scheme for the remotely manufacturing process. The remote-manufacturing process is continuously monitored and a crucial process is captured by CCD Camera. The captured image is compressed by neural network and JPEG, and it is sent directly to the assembly plant for incoming inspection. Massive image data require broadband channel to transmit them to remote distance, but sender is able to transmit them to receiver in use common channel by compressing massive image data in the high ratio. After the receiver reconstructs the compressed image to be transmitted, the reconstructed image is also directly used for automatic inspection of the process. The Experimental results show that the proposed inspection mechanism could be effectively implemented for real applications.

• 한국경영과학회:학술대회논문집
• /
• 대한산업공학회/한국경영과학회 2000년도 춘계공동학술대회 논문집
• /
• pp.135-138
• /
• 2000

This paper is intended to propose what manufacturing configuration (manufacturing planning and shop floor control) is suitable for the tire industry. Basically tire-manufacturing process is mixed-products, parallel-disconnected-flow-shop. Both throughput time and cycle tine are very short, the variety of tires is very high, the setup time is long, shop floor data reporting requirements is high, and there are many equipments and people working. And with no exception, tire industry also now confronts increasing requirements of delivery conformance with the above peculiar characteristics of tire manufacturing and changing market environments, this paper suggests, weekly master scheduling with no MRP is desirable and traditional kanban is right selection for shop floor control/scheduling. This paper describes why this configuration should be, using the manufacturing engineering principles and some new insights like four primitives of parallel flow shop. Generally known that shop with high parallel-product-mix and long setup time isn't good candidate for kanban. The four primitives of parallel flow shop explain why kanban is also useful scheduling technique in that environment.

• 한국콘텐츠학회논문지
• /
• 제17권7호
• /
• pp.182-192
• /
• 2017

제조 산업은 국가 경제 성장의 원동력으로 그 중요성이 부각되고 있다. 이에 따라 제조 공정상에서 생성되는 제조 데이터 분석의 중요성 또한 조명 받고 있다. 본 논문에서는 PCB(Printed Circuit Board) 제조 공정에서 발생한 로그 데이터를 분석하여 PCB 상에서 빈번하게 발생하는 고장 영역에 대해서 작업자가 고장 영역을 직접 눈으로 볼 수 있도록 시각화하는 방법을 제안한다. 우선 고장 영역을 파악하기 위해서 PCB 공정 데이터 집합에 K-means, DB-SCAN 클러스터링 알고리즘을 적용하여 군집화 하였고, 두 알고리즘 중 더 정확한 고장 영역을 도출하는지 비교하였다. 또한 MVC(Model-View-Controller) 구조 시스템을 개발하여 실제 PCB 이미지 상에 클러스터링 결과를 출력하는 것으로 실제 고장영역을 눈으로 확인할 수 있도록 시각화하였다.

• 한국안광학회지
• /
• 제12권4호
• /
• pp.87-90
• /
• 2007

polycarbonate(PC)용 안경렌즈를 생산하는데 사용되는 몰드(mold)를 가공하기 위해서는 코아(core)의 가공이 필요하다. 코아의 가공은 Diamond Turning Machine(DTM) 또는 Computer Numerical Control(CNC) 선반으로 이루어지는데 이러한 장비의 운용을 위해서는 렌즈 코아 형상에 대한 수치 데이터가 필요하게 된다. 이에 우리는 렌즈 코아 형상에 대한 수치 데이터를 산출하는 프로그램을 개발하였다. 프로그램은 일반 비구면 식의 계수를 사용하여 수치 데이터를 산출할 수 있도록 개발하였으며, 형상 그래프를 보여줄 수 있도록 하였고, 필요한 수치 데이터 파일을 저장할 수 있도록 하였다.

• Advances in nano research
• /
• 제17권2호
• /
• pp.109-124
• /
• 2024

In the era of smart manufacturing, precise prediction of springback-a common issue in ultra-thin sheet metal forming- and forming limits are critical for ensuring high-quality production and minimizing waste. This paper presents a novel approach that leverages the Internet of Things (IoT) and Artificial Neural Networks (ANN) to enhance springback and forming limits prediction accuracy. By integrating IoT-enabled sensors and devices, real-time data on material properties, forming conditions, and environmental factors are collected and transmitted to a central processing unit. This data serves as the input for an ANN model, which is trained with crystal plasticity simulations and experimental data to predict springback with high precision. Our proposed system not only provides continuous monitoring and adaptive learning capabilities but also facilitates real-time decision-making in manufacturing processes. Experimental results demonstrate significant improvements in prediction accuracy compared to traditional methods, highlighting the potential of IoT and ANN integration in advancing smart manufacturing. This approach promises to revolutionize quality control and operational efficiency in the industry, paving the way for more intelligent and responsive manufacturing systems.