Combinatory Prediction Analysis Of Vessel Accidents Based On Time Weight

The possibilities of vessel accidents should be lower and lower with the technology developing and more and more advanced equipment installed.However the actual vessel accidents in recent years remain at a relative high level;the investigations show the causes of high accidents are various mismanagements.Thus,it is very necessary to strengthen safety management on vessels.The Fujian coastal waters,which is located in Taiwan Strait,is an important seaway,thus it is significant to ensure the transportation safety of the waters,provide this,the significances of investigation on vessel accidents is evident and necessary.The predictions on trend and causes of vessel accidents in aforementioned coastal waters in a certain future period are made on basis of analysis on development law of vessel accidents incurred in past years,to positively established actions to the causes so as to avoid vessel accidents caused by same reason and reduce number of vessel accidents as far as possible,to eventually guarantee a safe seaways for vessels.The qualitative and quantitative are employed in combination in the study since the characteristics of integrated randomness,complexity and etc.of vessel accidents.The main causes of vessel accidents in Fujian coastal waters are concluded in this paper by carrying out statistic analysis on collected and organized vessel accidents form all marine bureaus of Fujian province;analyzing water features of hydro meteorology,track distribution and fishing vessel distribution,etc.;and in conjunction with typical accidents cases.After analyzing data characteristics of vessel accidents incurred in Fujian coastal waters within 2001~2014,and reference to methods used in other related studies in domestic and aboard,and conjunctions of the features and laws of the vessel accidents incurred in studied coastal waters,the GM(1,1)prediction model,ARIMA Time Series Forecasting Model and BP neural network predictive model,of which are incorporated in grey theory,are confirmed to be quite suitable for prediction on vessel accidents.The variations of vessel accident data could be effectively analyzed with results of single prediction on vessel accidents in Fujian coastal waters.The time-grouping forecast models for vessel accidents data of Fujian coastal waters are intendedly established to further improve predication accuracy of the model;and the satisfied conditions for time weight is given,and the time weight function is built,and the formula for calculating weight of time groupis derived.The demonstration indicates that the combinatory prediction model based on time weight possesses not only higher precision,when compared to a single predication method,but also better general applicability.At last,the suggestions on strengthening safety management of Fujian coastal waters are proposed to support scientific and effective decision basis for marine bureaus on controlling quantity of vessel accidents in time.