본 연구는 AI-Hub의 ‘방송콘텐츠 한국어-중국어 번역 말뭉치’ 데이터를 활용하여 네이버 파파고의 번역 품질을 평가하고 이를 시각화하는 것을 목표로 한다. 먼저 BL EU 점수를 통해 파파고 번역 품질을 정량적으로 분석하고, 다양한 시각화 기법을 활용하여 번역 품질의 패턴을 파악하였다. 연구 결과에 따르면, 파파고 번역은 짧고 단순한 문장에서 높은 성능을 보였으나, 문장이 길고 복잡해질수록 번역 품질이 저 하되는 경향이 나타났다. 특히, 구어체 표현이나 문화적 맥락이 필요한 표현에서는 번역 오류가 두드러지게 나타났다. 본 연구는 데이터 시각화를 통해 번역 품질 평가 의 이해를 돕고, 파파고 번역의 강점과 약점을 명확히 제시하였다. 향후 연구에서는 다양한 평가 지표와 데이터셋을 활용하여 번역 품질을 더욱 다각적으로 평가하고, 파파고 번역의 한계점을 개선 방안을 제시할 예정이다.

우리나라는 삼면이 바다로 이루어져 있고, 이에 따라 많은 해양 시설로 인한 위험유해물질이 배출되고 있으나, 배출관리 및 규제 시스템이 미비한 상황이다. 따라서, 위험유해물질(HNS) 관리를 위하여 효율적으로 데이터를 수집할 수 있는 시스템이 필요하 다. 본 연구에서는 HNS 데이터를 효율적으로 관리 및 저장하기 위한 데이터 표준화 시스템을 설계하고 이의 표준화 방안을 제시하고 자 한다.

In this study, we focus on the improvement of data quality transmitted from a weather buoy that guides a route of ships. The buoy has an Internet-of-Thing (IoT) including sensors to collect meteorological data and the buoy’s status, and it also has a wireless communication device to send them to the central database in a ground control center and ships nearby. The time interval of data collected by the sensor is irregular, and fault data is often detected. Therefore, this study provides a framework to improve data quality using machine learning models. The normal data pattern is trained by machine learning models, and the trained models detect the fault data from the collected data set of the sensor and adjust them. For determining fault data, interquartile range (IQR) removes the value outside the outlier, and an NGBoost algorithm removes the data above the upper bound and below the lower bound. The removed data is interpolated using NGBoost or long-short term memory (LSTM) algorithm. The performance of the suggested process is evaluated by actual weather buoy data from Korea to improve the quality of ‘AIR_TEMPERATURE’ data by using other data from the same buoy. The performance of our proposed framework has been validated through computational experiments based on real-world data, confirming its suitability for practical applications in real- world scenarios.

사출성형공정은 열가소성 수지를 가열하여 유동상태로 만들어 금형의 공동부에 가압 주입한 후에 금형 내에서 냉각시키는 공정으로, 금형의 공동모양과 동일한 제품을 만드는 방법이다. 대량생산이 가능하고, 복잡한 모양이 가능한 공정으로, 수지온도, 금형온도, 사출속도, 압력 등 다양한 요소들이 제품의 품질에 영향을 미친다. 제조현장에서 수집되는 데이터는 양품과 관련된 데이터는 많은 반면, 불량품과 관련된 데이터는 적어서 데이터불균형이 심각하다. 이러한 데이터불균형을 효율적으로 해결하기 위하여 언더샘플링, 오버샘플링, 복합샘플링 등이 적용되고 있다. 본 연구에서는 랜덤오버샘플링(ROS), 소수 클래스 오버 샘플링(SMOTE), ADASTN 등의 소수클래스의 데이터를 다수클래스만큼 증폭시키는 오버샘플링 기법을 활용하고, 데이터마이닝 기법을 활용하여 품질예측을 하고자 한다.

Ball stud parts are manufactured by a cold forging process, and fastening with other parts is secured through a head part cutting process. In order to improve process quality, stabilization of the forging quality of the head is given priority. To this end, in this study, a predictive model was developed for the purpose of improving forging quality. The prediction accuracy of the model based on 450 data sets acquired from the manufacturing site was low. As a result of gradually multiplying the data set based on FE simulation, it was expected that it would be possible to develop a predictive model with an accuracy of about 95%. It is essential to build automated labeling of forging load and dimensional data at manufacturing sites, and to apply a refinement algorithm for filtering data sets. Finally, in order to optimize the ball stud manufacturing process, it is necessary to develop a quality prediction model linked to the forging and cutting processes.

In this study, as part of the paradigm shift for manufacturing innovation, data from the multi-stage cold forging process was collected and based on this, a big data analysis technique was introduced to examine the possibility of quality prediction. In order for the analysis algorithm to be applied, the data collection infrastructure corresponding to the independent variable affecting the quality was built first. Similarly, an infrastructure for collecting data corresponding to the dependent variable was also built. In addition, a data set was created in the form of an independent variable-dependent variable, and the prediction accuracy of the quality prediction model according to the traditional statistical analysis and the tree-based regression model corresponding to the big data analysis technique was compared and analyzed. Lastly, the necessity of changing the manufacturing environment for the use of big data analysis in the manufacturing process was added.

With the recent development of manufacturing technology and the diversification of consumer needs, not only the process and quality control of production have become more complicated but also the kinds of information that manufacturing facilities provide the user about process have been diversified. Therefore the importance of big data analysis also has been raised. However, most small and medium enterprises (SMEs) lack the systematic infrastructure of big data management and analysis. In particular, due to the nature of domestic manufacturing companies that rely on foreign manufacturers for most of their manufacturing facilities, the need for their own data analysis and manufacturing support applications is increasing and research has been conducted in Korea. This study proposes integrated analysis platform for process and quality analysis, considering manufacturing big data database (DB) and data characteristics. The platform is implemented in two versions, Web and C/S, to enhance accessibility which perform template based quality analysis and real-time monitoring. The user can upload data from their local PC or DB and run analysis by combining single analysis module in template in a way they want since the platform is not optimized for a particular manufacturing process. Also Java and R are used as the development language for ease of system supplementation. It is expected that the platform will be available at a low price and evolve the ability of quality analysis in SMEs.

우리 해군은 함정 제품모델(Naval Ship Product Model, NSPM)을 중심으로 하는 협업 제품개발 환경 구축을 통해 설계 데이터의 재사용성과 M&S의 활용도를 높이고 있다. 그 결과 설계결과의 신뢰성이 높아지고 있으며, 이를 활용한 운영, 건조 시뮬레이션을 통해 소요군의 요구사항이 면밀히 반영되도록 하는 연구도 진행 중에 있다. 이에 따라 설계 데이터의 DB 구축 및 그 품질에 대한 중요성이 부각되고 있으나, 그와 관련된 연구는 초기상태에 머물러 있는 실정이다. 본 연구에서는 저자들의 선행연구 결과인 함정 제품모델의 품질검증 방법론에 기초하여 함정 제품모델을 구성하고 있는 형상요소의 구체적인 품질검증 방안과 이를 자동화하기 위한 연구를 수행하였다. 함정 제품모델 데이터 중 상대적으로 중요하며, 검증과정에 많은 시간이 소요되고 있는 선각모델을 사례연구 대상으로 정의하였으며, 자동차산업에서 사용되고 있는 제품데이터품질(Product Data Quality, PDQ)을 형상검증 기준으로 활용하였다. 최종적으로 선각모델의 형상 품질검증을 위한 기준과 알고리즘을 제안하였으며 이에 기반 한 초기 시스템을 개발하였다.

Data mining is the process of finding and analyzing data from a big database and summarizing it into useful information for a decision-making. A variety of data mining techniques have been being used for wide range of industries. One application of those is especially so for gathering meaningful information from process data in manufacturing factories for quality improvement. The purpose of this paper is to provide a methodology to improve manufacturing quality of fuel tanks which are auto-parts. The methodology is to analyse influential attributes and establish a model for optimal manufacturing condition of fuel tanks to improve the quality using decision tree, association rule, and feature selection.

This paper proposes the Data Architecture implementation process and the suggestions for automation which will enable the analyzing and measuring of the data structure as well. We verified the standardization of data that proposes a baseline for minimum quality, communications with stakeholder within an information system, minimization of application changes, and the effect of data values on quality improvement. The baseline is tested through the measurement and assessment of the quality indicator of a data structure as well as a quality control process deduced during the implementation.

The paper proposes the misuse types of statistical quality tools according to the kind of data and the number of population in DMAIC process of six sigma. The result presented in this paper can be extended to the QC story 15 steps of QC circle. The study also provides the improvement methods about control chart, measurement system analysis, statistical difference, and practical equivalence.