Research On Marine Collision Accident Risk Based On Ontology Model

Great efforts have been made in the development of the study on risk factors of maritime collision accidents.However,the interoperability of theories in the field of maritime collision and the rate of knowledge reuse are regrettably low.The attention should be paid more on the how to conclude and share the actual abundant results and experiences.Main contributes of this paper are firstly combining Ontology and Bayesian model in the field of maritime accident security.Ontology model is established to sort out and formalize professional concepts,relationships and attributes involved in the collision field.Ontology Model-driven Bayesian model extends the probability of influencing factors,and be responsible for inference and prediction of maritime collision accident risk.There are three main conclusions as follow.Firstly,loading,accident time,gross tonnage,wind,ship type and judgment,are determined to be the most significant influencing factors of the accident degree in ship collision accidents.Then,four sequences of significant influencing factors of major accidents and critical accidents are obtained.Last,the influence contents of heavy weather on accident severity level are pointed out as follow.Wind has a great influence on accident severity compared with visibility.when wind does not reach 8 degree,it has no significant influence on the accident level.When the wind reaches grade 8or above,the occurrence rate of major accidents will be significantly increased and the occurrence rate of critical accidents will be reduced by 100%.In addition,minor accidents occurs most likely to happen in the situation of good wind situation and worst visibility,the situation of worst wind situation and good visibility has highest probability to occurs major accidents,critical accidents mostly occurs in situation of good wind situation and good visibility.This paper aims to build a maritime collision scenario model based on the combination of the Ontology model and the Bayesian model,which combines advantages of high expression ability and high transformability of Ontology model and the uncertain reasoning ability of Bayesian model.Coupling Ontology with Bayesian model,a collision accident risk reasoning model with high generality and strong reasoning ability is constructed.It not only has good expression ability and computer readability for knowledge storage,but also can realize the function of reasoning and prediction for accident.Meanwhile,its strong universality makes it applicable to the research in other maritime safety fields...