http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Fu Xingyu,Zhou Fengfeng,Peddireddy Dheeraj,Kang Zhengyang,Jun Martin Byung-Guk,Aggarwal Vaneet 한국CDE학회 2023 Journal of computational design and engineering Vol.10 No.3
In this work, we present a boundary oriented graph embedding (BOGE) approach for the graph neural network to assist in rapid design and digital prototyping. The cantilever beam problem has been solved as an example to validate its potential of providing physical field results and optimized designs using only 10 ms. Providing shortcuts for both boundary elements and local neighbor elements, the BOGE approach can embed unstructured mesh elements into the graph and performs an efficient regression on large-scale triangular-mesh-based finite element analysis (FEA) results, which cannot be realized by other machine-learning-based surrogate methods. It has the potential to serve as a surrogate model for other boundary value problems. Focusing on the cantilever beam problem, the BOGE approach with 3-layer DeepGCN model achieves the regression with mean square error (MSE) of 0.011 706 (2.41% mean absolute percentage error) for stress field prediction and 0.002 735 MSE (with 1.58% elements having error larger than 0.01) for topological optimization. The overall concept of the BOGE approach paves the way for a general and efficient deep-learning-based FEA simulator that will benefit both industry and Computer Aided Design (CAD) design-related areas.
Machine Learning for Object Recognition in Manufacturing Applications
Huitaek Yun,Eunseob Kim,Dong Min Kim,Hyung Wook Park,Martin Byung-Guk Jun 한국정밀공학회 2023 International Journal of Precision Engineering and Vol.24 No.4
Feature recognition and manufacturability analysis from computer-aided design (CAD) models are indispensable technologies for better decision making in manufacturing processes. It is important to transform the knowledge embedded within a CAD model to manufacturing instructions for companies to remain competitive as experienced baby-boomer experts are going to retire. Automatic feature recognition and computer-aided process planning have a long history in research, and recent developments regarding algorithms and computing power are bringing machine learning (ML) capability within reach of manufacturers. Feature recognition using ML has emerged as an alternative to conventional methods. This study reviews ML techniques to recognize objects, features, and construct process plans. It describes the potential for ML in object or feature recognition and offers insight into its implementation in various smart manufacturing applications. The study describes ML methods frequently used in manufacturing, with a brief introduction of underlying principles. After a review of conventional object recognition methods, the study discusses recent studies and outlooks on feature recognition and manufacturability analysis using ML.
Geometrical Simulation Model for Milling of Carbon Fiber Reinforced Polymers (CFRP)
Xingyu Fu,송경은,Dong Min Kim,Zhengyang Kang,Martin Byung-Guk Jun 한국정밀공학회 2022 International Journal of Precision Engineering and Vol.23 No.11
This paper proposes a geometrical 3D milling simulation algorithm for Carbon Fiber Reinforced Polymer (CFRP) milling. In this simulation model, milling tools are simplified into layers of circles while CFRP lami- nates are simplified into layers of Dexel lines, which can realize simulations for various complex milling conditions. Significant geometrical parameters, for ex-ample, cutting angle and cutting length, can be computed with high efficiency. With some geometry-related physical models, the machining results can be pre-dicted for the entire milling process. The effectiveness of this simulation model has been validated by the milling force prediction and the delamination pre-diction. The performance of this simulation model benefits industrial CFRP manufacturing and provides a new method for online or long-time-interval simulation of CFRP machining.
Chengwen Han,Kyeong Bin Kim,Seok-Woo Lee,Martin Byung-Guk Jun,Young Hun Jeong 한국정밀공학회 2021 International Journal of Precision Engineering and Vol.22 No.9
Currently, carbon fiber-reinforced plastic (CFRP) is a material with potential uses for various industries due to its excellent properties. However, the severe tool wear in its machining is an evitable problem because it deteriorates product quality and productivity at the same time. Timely replacement of the tool should be taken in advance to reduce the influences of this drawback. In sum, the accurate estimation of the tool wear plays a crucial role in CFRP machining. Therefore, in this study, a tool wear estimation in CFRP drilling is presented. The method used is based on discrete wavelet transformation (DWT) of the thrust force signal and an artificial neural network (ANN). Two valuable features related to tool wear are identified and then extracted using DWT. The tool wear is estimated by using an ANN with two features adapted from DWT. Consequently, the tool wear, especially flank wear, in CFRP drilling can be accurately estimated using the proposed method.