세계 해양산업은 자율운항선박 기술의 등장으로 급속도로 발전하고 있으며, 해양 데이터에서 파생된 인공지능 활용에 관한 관 심이 높아지고 있다. 다양한 기술 발전 중에서 선박 항로 군집화는 자율운항선박 상용화를 위한 중요한 기술로 부각되고 있다. 항로 군집 화를 통해 해상에서 선박 항로 패턴을 추출하여 가장 빠르고 안전한 항로를 최적화하고 충돌 방지 시스템의 개발에 기반이 된다. 항로 군 집화 알고리즘의 정확성과 효율성을 보장하기 위해 고품질의 잘 처리된 데이터가 필수적이다. 본 연구에서는 다양한 항로 군집화 방법 중 항로의 실제 형태와 특성을 정확히 반영할 수 있는 선박 항로 유사도 기반 군집화 방식에 주목하였다. 이러한 방식의 효율을 극대화하 기 위해 최적의 데이터 전처리 기술 조합을 구성하고자 한다. 구체적으로, 4가지의 선박 항로 간 유사도 측정법과 3가지의 차원 축소 방 법을 조합하여 연구를 진행하였다. 각 조합에 대해 k-means 군집 분석을 수행하고, 그 결과를 Silhouette Index를 통해 정량적으로 평가하여 최고 성능을 보이는 전처리 기법 조합을 도출하였다. 본 연구는 단순히 최적의 전처리 기법을 찾는 것에 그치지 않고, 광범위한 해양 데 이터에서 의미 있는 정보를 추출하는 과정의 중요성을 강조한다. 이는 4차 산업혁명 시대의 해양 및 해운 산업이 직면한 디지털 전환에 효과적으로 대응하기 위한 기초 연구로서 의의를 갖는다.

PURPOSES : In this study, a preliminary study on the optimal clustering techniques for the preprocessing of pavement management system (PMS) data was conducted using K-means and mean-shift techniques to improve the correlation between the dependent and independent variables of the pavement performance model. METHODS : The PMS data of Jeju Island was preprocessed using the K-means and mean-shift algorithms. In the case of the K-means method, the elbow method and silhouette score were used to determine the optimal number of clusters (K). Moreover, in the case of the mean-shift method, Scott’s rule of thumb and Silverman’s rule of thumb were used to determine the optimal cluster bandwidth. RESULTS : The optimal cluster sets were selected for the rut depth (RD), annual average daily traffic (AADT), and annual maximum temperature (AMT) for each clustering technique, and their similarities with the original data were investigated. Additionally, the correlation improvement between the dependent and independent variables were investigated by calculating the clustering score (CS). Consequently, the K-means method was selected as the optimal clustering technique for the preprocessing of PMS data. The K-means method improved the correlations of more variables with the dependent variable compared to the mean-shift method. The correlations of the variables related to high temperature—such as the annual temperature change, summer days, and heat wave days—were improved in the case wherein the AMT, a climate factor, was used as an independent variable in the K-means clustering method. CONCLUSIONS : The applicability of the clustering methods to preprocessing of PMS data was identified in this study. Improvements in the pavement performance prediction model developed using traditional statistical methods may be identified by developing a model using clustering techniques in a future study.

Recently, a study of prognosis and health management (PHM) was conducted to diagnose failure and predict the life of air craft engine parts using sensor data. PHM is a framework that provides individualized solutions for managing system health. This study predicted the remaining useful life (RUL) of aeroengine using degradation data collected by sensors provided by the IEEE 2008 PHM Conference Challenge. There are 218 engine sensor data that has initial wear and production deviations. It was difficult to determine the characteristics of the engine parts since the system and domain-specific information was not provided. Each engine has a different cycle, making it difficult to use time series models. Therefore, this analysis was performed using machine learning algorithms rather than statistical time series models. The machine learning algorithms used were a random forest, gradient boost tree analysis and XG boost. A sliding window was applied to develop RUL predictions. We compared model performance before and after applying the sliding window, and proposed a data preprocessing method to develop RUL predictions. The model was evaluated by R-square scores and root mean squares error (RMSE). It was shown that the XG boost model of the random split method using the sliding window preprocessing approach has the best predictive performance.

Recently, measurement monitoring is actively used for safety management of facilities. However, since the field measurement data contains many outliers, a preprocessing process is required for reliable behavior analysis of the data. In this paper, we present a detection method of time series outliers and its applications. And we propose the precaution for the preprocessing process.

Recently, measurement monitoring is actively used for safety management of facilities. However, since the field measurement data contains many outliers, a preprocessing process is required for reliable behavior analysis of the data. In this paper, we present a detection method of time series outliers and its applications. And we propose precautions for the preprocessing process.