http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Zahra Nourmohammadi,권동현,김인희 대한교통학회 2021 대한교통학회 학술대회지 Vol.85 No.-
The resulting increases in heavy ocean traffic flow have made maritime accidents a complex problem in most waterways. Thus, ocean accident analysis is one of the prime interests to enhance marine safety. For reducing accidents, it is important to identify the accident blackspots. Our dataset contains 12,329 numbers of records from 2014 to 2018. Case, Name, Jurisdiction, Types of Accident, Date, Time, Season, Latitude, Longitude, Region, Importance, Capacity, Territory, Purposes of Travel, Number of Casualty are the variables in our dataset. We benchmarked 5 clustering algorithms such as K-Means, DBSCAN, Spectral, Affinity Propagation, Agglomerative, and a grid-based clustering called CLIQUE to find the blackspots by finding the optimum parameters of each. The outcomes help ocean accident planners to pay more attention to the blackspots and more patrol ships can be employed by decision makers in certain areas to avoid risks.
해양사고 저감을 위한 위험지역 순찰 및 사후 골든타임 확보를 위한 빅데이터와 인공지능 기반의 최적화 서비스 개발
Zahra Nourmohammadi,Fatemeh Nourmohammadi,김인희 대한교통학회 2021 대한교통학회 학술대회지 Vol.85 No.-
Patrol ship Route Analysis and Optimization Model The demand for the shipping services and seaborn trade has increased in the past several years increased possibility of navigation accident, illegal fishing etc. Consequently, ship navigation safety has become a growing concern in maritime transportation. Patrol ships play a vital navigation safety enhancement in the marine environments. This study proposes a minimized patrol path in South Korea. Firstly, the model considered as a Travelling Salesman Problem (TSP) and solved with a heuristic algorithm (Simulated Annealing) and the result shows a good result. In the next step, we consider more constraints which could not be considered in the TSP model because suborn eliminations. Therefore, we proposed a new model to solve a linear programming considering all constraints as the constraints were important for us. Korea Maritime Accident Prediction Based on Clustering and a grid-based Classification Method Maritime Safety has become one the top global concerns recently. Maritime accidents are directly connected with human lives, environment, and economy. Particularly, shipping has long been regarded as a complex and high-risk activity due to the uncertainty and severe condition in the sea. The data is collected from 2014 to 2018 (a 4-year period) with total number of 12329 records. Types, Date and Time, Season, Location (Longitude and Latitude), Region, Capacity, Purpose of Travel, and total casualty are considered as variables. As preliminary analysis results, we found accident black spots using Four Clustering algorithms such as K-means, DBSCAN, Agglomerative Clustering which second one showed the best result. After making grids in three most dense areas, we predicted grids that illustrates the opproximate location of the accident using Logistic Regression, K-Nearest Neighbors, Decision Tree, Random Forest classification. Decision Tree (Entropy )showed the best accuracy for this Prediction for all the grid sizes and all the clusters. The outcomes will be helpful in guiding the management of maritime traffic safety to pay more attention to the area with higher danger of accident. Results of classification which presents new insights for accident prevention practice for maritime authorities.