Research On Forecast Of Ship Traffic Accidents Based On MGM(1, N) Model

In this paper, firstly we analysis the affecting factors about ship traffic accidents including people factors, vessel factors, environment factors, and management factors etc; after referring a lot of information, analyzing and summarizing the previous research, this paper confirms the relatively important factors on the basis of previous qualitative analysis such as expert investigation method and accident tree method and so on, then use the grey relational methods to do a further quantitative analysis and selection, because the ordinary gray relational method has the flaw of interrelatedness difference not obvious, this article used the improvement grey relational analysis method to overcome this shortcoming, finally get the more relative factors system concerning the ship traffic accidents, which pave the way for establishing the model of accidents forecast.Because the ship traffic accidents are the random things which are affected, constrained and the jointly developed factors, so establishing the traffic accidents model according to the factors conforms to the actual situation relatively. so after analyzing all the present ship traffic accidents forecast model, this paper uses multi-factor grey model MGM(1,n) for predicting ship traffic accident. Modeling in practice, due to the smoothness about the accident prediction date is not good enough, in order to improve the smoothness about the original data sequence, enhance the regularity of the original data, weaken its randomness, and improve prediction accuracy, this paper use power function transformation for data proceeding, and then use generated sequence to establish multi-variable grey predicting model, using actual date to test the model, after initially obtains the ships traffic accident forecast result, in order to obtain the further optimization of the forecast model and enhance the forecast precision of the model, finally do the residual analysis to the predicted model on the basis of the number of residuals, then establish residual GM(1,1) model, and do the residual correction to achieve the model optimization and improve the further prediction accuracy. The results show that the accidents forecast model reveals the rule of ship traffic accident and associated factors within a certain margin of error.