| Once marine traffic accident occurs, it may cause severe casualties, property loss and environment pollution which have serious impact on the development of economic and society. So it is of great importance to make a reasonable and effective prediction to the regular pattern of marine traffic accident in order to take appropriate response measures. Although marine traffic accident has many characteristics, such as various influent-factors, irregular, randomness, higher-dimension nonlinear, data missing and so on due to the fact that it is influenced by natural conditions, human factors, shipping status and so on, its internal laws can be found through the analysis and research of historical data, so as to make reasonable forecast of the number of marine traffic accident within a certain range of area in the near future. This paper mainly predicts marine traffic accident discipline in Ningbo area based on historical data of marine traffic accident in Ningbo area with different prediction models. In order to give fully play to the advantages of these forest methods and avoid their shortcomings, the final result is the weighted average value of different forecast models. First this paper makes comprehensive analyze of the natural conditions, water conditions, ship navigation, etc. of Ningbo area that have potential influence on marine traffic accident. Then the requirement of marine traffic accident is researched in human factors, environmental factors, ship factors and management factors, so as to take effective measures to avoid it. According to the marine traffic accident data of Ningbo area, the accident-prone period and waters can be identified, so Maritime Safety Administration(MSA) can make targeted regulation. At last, the predictions of marine traffic accident in 2009 is obtained by linear regression analysis, exponential smoothing method and grey theory(GM(1,1)), and the number of marine traffic accident that may occur in 2010 is predicted using proper prediction method.
|
PDF Full Text Request