Risk Research Of Ocean Ship Collision Based On Historical Data Analysis

Among many modes of transportation,waterway transportation is generally accepted and is developing rapidly,because the cost of it is cheap and amount of cargo is large.But waterway transportation is easily influenced by natural environment,poor flexibility.Ship collision accident rate is prevailing and may lead to property loss,loss of life and personal injury,environmental disruption.In order to reduce the risk of ship collision and control the principal factor,this paper take ship collision as research object,ship collision of probability and consequence are analyzed and calculated,the details are as follows:(1)Ocean ship collision accident reports are collected,causation factors(such as human factor,management factor,equipment factor and environment factor),ship characters(such as age,weight,type,and location of collision,meeting situation,number of people and speed),number of injury,death and damage of ship in accidents.The causation factors is classified and numbered.The characters of ship is classified according to standard of IMO and China.Number of injury,death and damage of ship are taken as the index to assessment the consequence of ocean ship collision.Equivalent consequence is used to quantitate the consequence of ship collison accident.(2)The way to generate candidate frequent item set has been replaced to reduce the workload of this arithmetic to improve the arithmetic speed of traditional Aprioro arithmetic.The analysis about the causation factors is carried out based on the improvement algorithm.It is found that the mean factors which influence ocean ship collision significantly are improper outlook,misestimation of collision risk,operated improperly,not take action immediately and communicate insufficiently in order.(3)Decision tree is built based on the CHAID algorithm to analyze the relationship between ship charactors and Number of injury,death,damage of ship and equivalent consequence.It is found that type of ship,location of collision and meeting situation.Small ship,passenger ship and general cargo ship is selected to find the charactors and its' classification.(4)BP neural network is build,in which ship charactors are input and consequence of collision is output.After training and testing,the accuracy rate of this model is more than 90%.Permutation and combination of ship charactors are carried out for all kinds of situation,which are possible.The ship charactors composite modes above are put into BP neural network for many times and the consequences will be calculated.The consequence of general,serious and very serious about small ship is 0.602778,0.322222,0.075,passenger ship is 0.9963,0,0.0037,general cargo ship is 0.9333,0.025,0.00417.The consequence of ships entering and leaving port can be predicted basen on the BP neural network above.(5)Probability of different kinds of ships is calculated based on the statistical data.The risk of the three kinds of ships is calculated based on the probability and consequence above.Risk of general,serious and very serious ocean ship collision about small ship is 2.321483E-03,1.240976E-03,2.888480E-04,passenger ship is 1.82828E-05,0,6.78975E-08,general cargo ship is 8.99076E-05?2.40833E-06?4.01709E-07.It is easy to find that collision risk of small ship is highest.In general,the risk of general collision is highest,very serious collision is lowest.