Applications Of Cloud Theory And Data Mining In Marine Safety Analysis

Marine safety has been a core subject in marine accidents studies, because it is connected with transport safety, shipping efficiency, and national economy and people's lives and properties loss prevention closely. Since the beginning of the 21st century, the rising revolution of science and technology and the globalization of trade have led to a rapid increase in the requirement of marine transportation and vessel movements. Yet the probability of maritime accidents and hazardous incidents is increased simultaneously, followed the well-developed marine transportation by the economic growth, which has made the requests of marine security higher and higher. Therefore, this thesis attempts to solve the major issues in marine safety and discusses the corresponding theory and technique of cloud model and data mining and their applications in marine safety analysis. The contents and results are listed, as follows.Firstly, the thesis analyzes the current state of related fields of marine safety, and then three major issues of maritime traffic safety are presented.Secondly, due to the lack of necessary data for maritime traffic safety planning, the elementary database of maritime traffic accidents is developed by fully investigated and sufficient survey and analysis based on a large number of maritime accidents in different sea areas and inland waterways, which is available to planning's research and making.Thirdly, to meet the needs of data mining of data analysis, the elementary marine traffic data is considered as the data source of data mining, which is necessary to data processing to maritime accidents, considering the self-characteristics of maritime accident data. Then the algorithm of association rules is introduced to explore the knowledge accordance with actual regulars so as to find out the preferential information behind a great quantity of maritime casualties. And accident pattern mining of entire element, preferential element pattern mining and pattern mining based on consequence are carried out respectively. Consequently, the corresponding rules are obtained, which is beneficial to understanding the cause and tendency of maritime accidents. That provides a combination of qualitative and quantitative basis for decision making on planning of traffic safety and security on the sea. Fourthly, aiming to process both fuzziness and randomness in the complex marine traffic safety, especially the human element as the most important cause in maritime casualties, a novel cloud model based assessment paradigm is proposed, where various uncertainties in the system are well kept and modeled by cloud theory. The synthetic assessment method based on cloud model is much more suitable to attribute factors, which are expressed by natural language values. And it is put to seafarers' competency evaluation. Simulation research shows that the presented model is more intuitive, flexible, reliable and consistent with human thinking.Lastly, the conclusion is made, and the problems for further study are reviewed.