Abstract: In the field of intelligent traffic, a reasonable model for risk management and decision making is a powerful tool for utilizing limited surveillance resources and mitigating the risks of marine accidents. The existing models place emphasis on direct scoring for each individual factor of ship risk profile and ranking by ship’s cumulative scores, which underrepresents dynamic states of ship risk profile and sequence of distribution of factors. In the meantime, the models available at present give little consideration of the human factor and tend to give high scores to particular types of ships with low risks which distorts the true risk level of such ships. The paper develops a grading model for ship risk based on FCM and PROMETHEE-Ⅱ. Firstly, the degrees of domain at all factors are obtained through PROMETHEE-Ⅱ. Secondly, cluster analysis through FCM is made to embrace the data distribution characteristics and achieve the aims of dynamic management and classification. In order to ascertain the effectiveness of the proposed model, the data of 400 ships sampled in the case study is mainly from PSC MOU, authentic marine safety reports, shipxy.com and investigations. It is demonstrated that there is higher fitting degree between classification results of the model suggested and evaluation results of ships with existing recognized ship quality evaluation systems, and presents a feature of dynamic sorting. The model built in this paper can be applied to data processing of integrated quality management system for vessels, which may assist the officials to identify the ships with relative high risk profile. Key words: ship risk, FCM, PROMETHEE-Ⅱ

CICTP