• 반도체디스플레이기술학회지
• /
• 제18권3호
• /
• pp.25-30
• /
• 2019

In this paper, we have analyzed manufacturing problems of the scaffold with pores using FDM 3D printer and PLGA. We suggested the solutions using 3SC practical TRIZ. We selected the final solution used machine learning. We reduced number of experiments using most influential factor after analysis print factors. We printed the scaffold and measured pore size. We created the regression model using python tensorflow. The print condition data of measured pore size was used as training data. We predicted the pore size of printed condition using regression model. We printed the scaffold using the predicted the print condition data. We quantitatively compare the predicted scaffold pore size data and the measured scaffold pore size data. We got satisfactory result.

• 한국철도학회논문집
• /
• 제7권1호
• /
• pp.1-8
• /
• 2004

A newly-built inverter has to undergo a series of stress tests in the final stage of production line. This can be achieved by connecting it to a dynamometer consisting of a three-phase machine joined by a rigid shaft to a DC load machine. The latter is controlled to create some specific load characteristic needed for the test. In this paper a test method is proposed, in which no mechanical equipment is needed. The suggested test stand consists only of a inverter to be tested and a simulator converter. Both devices are connected back-to-back on the AC-side via smoothing reactors. The simulator operates in real-time as an equivalent load circuit, so that the device under test will only notice the behaviour of a three-phase machine under consideration of the load. In oder to prove rightness of the suggested test method, the simulation and actural experiment carried out emulation for a 2.2kW induction motor.

• Journal of information and communication convergence engineering
• /
• 제17권1호
• /
• pp.8-13
• /
• 2019

Smart Farming has been regarded as an important application in information and communications technology (ICT) fields. Selecting crops for cultivation at the pre-production stage is critical for agricultural producers' final profits because over-production and under-production may result in uncountable losses, and it is necessary to predict crop production to prevent these losses. The ITU-T Recommendation for Smart Farming (Y.4450/Y.2238) defines plan/production consultation service at the pre-production stage; this type of service must trace crop production in a predictive way. Several research papers present that machine learning technology can be applied to predict crop production after related data are learned, but these technologies have little to do with standardized ICT services. This paper clarifies the relationship between agricultural consultation services and predicting crop production. A prediction scheme is proposed, and the results confirm the usability and superiority of machine learning for predicting crop production.

• 산업경영시스템학회지
• /
• 제46권4호
• /
• pp.181-189
• /
• 2023

In this paper, we present a case study of developing MVIS (Machine Vision Inspection System) designed for exterior quality inspection of stamping dies used in the production of automotive exterior components in a small to medium-sized factory. While the primary processes within the factory, including machining, transportation, and loading, have been automated using PLCs, CNC machines, and robots, the final quality inspection process still relies on manual labor. We implement the MVIS with general-purpose industrial cameras and Python-based open-source libraries and frameworks for rapid and low-cost development. The MVIS can play a major role on improving throughput and lead time of stamping dies. Furthermore, the processed inspection images can be leveraged for future process monitoring and improvement by applying deep learning techniques.

• Structural Engineering and Mechanics
• /
• 제88권6호
• /
• pp.535-549
• /
• 2023

Deep extreme learning machine (DELM) and multi-verse optimization algorithms (MVO) are hybridized for designing an optimal and adaptive control framework for uncertain buildings. In this approach, first, a robust model predictive control (RMPC) scheme is developed to handle the problem uncertainty. The optimality and adaptivity of the proposed controller are provided by the optimal determination of the tunning weights of the linear programming (LP) cost function for clustered external loads using the MVO. The final control policy is achieved by collecting the clustered data and training them by DELM. The efficiency of the introduced control scheme is demonstrated by the numerical simulation of a ten-story benchmark building subjected to earthquake excitations. The results represent the capability of the proposed framework compared to robust MPC (RMPC), conventional MPC (CMPC), and conventional DELM algorithms in structural motion control.

• Journal of Biosystems Engineering
• /
• 제38권1호
• /
• pp.9-17
• /
• 2013

Purpose: The goal of this study was to develop a methodology for the demand forecast of tractor, riding type rice transplanter and combine harvester using an ARIMA (autoregressive integrated moving average) model, one of time series analysis methods, and to forecast their demands from 2012 to 2021 in South Korea. Methods: To forecast the demands of three kinds of machines, ARIMA models were constructed by following three stages; identification, estimation and diagnose. Time series used were supply and stock of each machine and the analysis tool was SAS 9.2 for Windows XP. Results: Six final models, supply based ones and stock based ones for each machine, were constructed from 32 tentative models identified by examining the ACF (autocorrelation function) plots and the PACF (partial autocorrelation function) plots. All demand series forecasted by the final models showed increasing trends and fluctuations with two-year period. Conclusions: Some forecast results of this study are not applicable immediately due to periodic fluctuation and large variation. However, it can be advanced by incorporating treatment of outliers or combining with another forecast methods.

• Journal of Biosystems Engineering
• /
• 제32권2호
• /
• pp.84-90
• /
• 2007

This study was conducted to design and manufacture a final prototype of garlic separator and to evaluate its performance. The performance of garlic separation was compared with manual separation. The final prototype for garlic separation consists of bucked-elevator device for automatic feed of garlic, rotating cone typed device, blower, and power transmission device. The optimal condition of outlet clearance was 19 mm and in this clearance, the proportions of fragment garlic separated in the large quality of Namdo garlic and all quality of Uiseong garlic were above 95% and above 85%, respectively. All proportion of damaged garlic was below 5% for all variety and quality. The garlic separation capacities of this developed machine were 310 kg/h for Namdo garlic and 293.6 kg/h for Uiseong garlic in the large quality. Capacities of final prototype compared with human being were $12.9{\sim}19.6$ times for Namdo and $24.2{\sim}31.7$ times Uiseong garlic, respectively.

• 한국게임학회 논문지
• /
• 제21권4호
• /
• pp.3-12
• /
• 2021

본 연구에서는 게임 초반 10분의 데이터를 이용하여 리그오브레전드 게임의 최종승패를 랭크별로 예측하고, 구축된 승패예측 모형으로부터 변수중요도를 추출하여 승리를 위한 초반 게임운영의 방향성을 알아보았다. 그 결과 모든 랭크에서 70% 이상의 정확도로 승패를 예측할 수 있었다. 이는 경기 양상이 대부분 뒤집히지 않고 최종승패로 이어지는 것을 의미하며, 이러한 경향성은 상위 랭크로 갈수록 더욱 강하게 나타났다. 랭크와 무관하게 킬(데스)가 초반 게임에서 최종승패에 가장 큰 영향을 미치는 요소로 나타났으나, 일부 변수는 랭크에 따라 중요도 순위가 변화하였고 이는 유저가 속한 랭크에 따라 승리에 효과적인 초반 전략방향에 차이가 있음을 시사한다.

• 한국컴퓨터정보학회논문지
• /
• 제26권2호
• /
• pp.61-68
• /
• 2021

최근 악성코드와 정상 바이너리를 분류하기 위해 기계학습을 이용하는 기술이 다양하게 연구되고 있다. 효과적인 기계학습을 위해서는 악성코드와 정상 바이너리를 식별하기 위한 Feature를 잘 추출하는 것이 무엇보다 중요하다. 본 논문에서는 재귀적인 방법을 이용하여 기계학습에 활용하기 위한 Feature 추출 방법인 RFA(Recursive Feature Addition) 제안한다. 제안하는 방법은 기계학습의 성능을 극대화 하기 위해 개별 Feature를 대상으로 재귀적인 방법을 사용하여 최종 Feature Set을 선정한다. 세부적으로는 매 단계마다 개별 Feature 중 최고성능을 내는 Feature를 추출하여, 추출한 Feature를 결합하는 방법을 사용한다. 제안하는 방법을 활용하여 Decision tree, SVM, Random forest, KNN등의 기계학습 알고리즘에 적용한 결과 단계가 지속될수록 기계학습의 성능이 향상되는 것을 검증하였다.

• 한국분말재료학회지
• /
• 제11권4호
• /
• pp.308-313
• /
• 2004

An experimental study on the combustion of superfine aluminum powders (average particle diameter, a$_{s}$: ∼0.1 ${\mu}{\textrm}{m}$) in air is reported. The formation of aluminum nitride during the combustion of aluminum in air and the influence of the combustion scenario on the structures and compositions of the final products are in the focus of this study. The experiments were conducted in an air (pressure: 1 atm). Superfine aluminum powders were produced by the wire electrical explosion method. Such superfine aluminum powder is stable in air but once ignited it can burn in a self-sustaining way due to its low bulk: density (∼0.1 g/㎤) and a low thermal conductivity. During combustion, the temperature and radiation were measured and the actual burning process was recorded by a video camera. Scanning electron microscopy (SEM), X-ray diffraction (XRD) and chemical analysis were performed on the both initial powders and final products. It was found that the powders, ignited by local heating, burned in a two-stage self-propagating regime. The products of the first stage consisted of unreacted aluminum (-70 mass %) and amorphous oxides with traces of AlN. After the second stage the AlN content exceeded 50 mass % and the residual Al content decreased to ∼10 mass %. A qualitative discussion is given on the kinetic limitation for AlN oxidation due to rapid condensation and encapsulation of gaseous AlN.N.