http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
박준현(Jun-Hyun Park),장민국(Min-Kook Jang),이강희(Gang-Hui Lee),오은경(Eun-Kyung Oh),허성우(Sung-Woo Hur) 한국정보기술학회 2016 한국정보기술학회논문지 Vol.14 No.11
As the size of a vessel is getting bigger, the speed of that is getting higher, and the most controling mechanism is automated, the volume of marine cargo and the number of vessels held by a company increases continually. So, the role of ship management industry becomes more important. A sudden failure of major equipments of a vessel causes a great economical damage to the vessel management company, hence an algorithm which can predict and prevent failures in advance is required. Most of existing algorithms for the requirement have a limit in predicting failures in advance because they usually use simple method like setting the threshold values of sensors or examining the correlation between sensors. In this paper, we present an effective algorithm to detect engine failure, which reduces a large amount of sensor data into smaller set of useful data by analysing correlation among them and which detects defect data by regression analysis. Results obtained by simulation using the real data generated by a vessel, which belongs to H marine company, and the failure report, our proposed algorithm is proved to be effective to predict engine failure in advance.
박준현(Jun-Hyun Park),오은경(Eun-Kyung Oh),장민국(Min-Kook Jang),서영우(Young-Woo Seo),허성우(Sung-Woo Hur) 한국정보기술학회 2017 한국정보기술학회논문지 Vol.15 No.11
Due to the automation and speeding up of large vessels, the vessels operation schedule has been increased, and as a result, the docking time has become shorter, resulting in a shortage of maintenance time. This increases the probability of vessel failure. Vessel failure at a ocean causes serious consequences in many respects, so techniques that can predict the failure is necessary. A system to predict vessel failure using linear regression in real time has been suggested in previous paper. In this paper, we propose techniques for data-cleansing, automatic prediction interval setting, spline linear regression, sliding window, and for showing user-oriented time series graph to improve the accuracy of error detection. Results obtained by simulation using the real data and the failure report generated by a vessel, which belongs to H marine company, prove that the improved algorithm is more effective in predicting engine failure in advance compared to the methods used in previous paper.