Risk Control Of LNG Ship Loading And Unloading Leakage Based On Bayesian Network

With the substantial increase in the global import and export volume of liquefied natural gas(LNG)and the continuous increase in the number of orders of LNG ships,it is expected that the loading and unloading operations of LNG will be more frequent in the future.Therefore,the research related to the safety of LNG ships has become a top priority.The loading and unloading operation of LNG ship is a very importance,and the process is complex,lasts for a long time and involves many people.If there is a mistake in a certain link,it will lead to LNG leakage accidents and explosion accidents,resulting in casualties and economic losses.The risk control of LNG ship loading and unloading operation can greatly ensure the overall safety of LNG ship transportation.Therefore,the risk control is of great significance to ensure the safety and interests of personnel,ships and the environment.In order to evaluate the leakage risk of LNG ship loading and unloading operation,this thesis first analyzes and extracts the risk factors existing in LNG ship loading and unloading operation,and extracts the risk factors classified as human factors,ship factors and environmental factors from historical accidents and loading and unloading process.Based on Bayesian network(BN),the risk nodes are obtained,and the historical accident data are adjusted according to the format required by BN.Secondly,considering the lack of historical accident data,combined with kernel density estimation(KDE),the small sample data is expanded on the basis of retaining the original characteristics and distribution of the data.Finally,BN learning will be carried out on the expanded data to obtain the relationship and influence degree of each risk node,so as to improve the ship and port’s understanding of LNG ship handling risk and assist its decision-making.The innovative results of this study:(1)The leakage risk model of LNG ship loading and unloading is established by BN.(2)According to the characteristics of small samples of LNG ship accident data,the method of nuclear density estimation is applied to expand.(3)BN structure learning is carried out in a data-driven way.(4)Graphically represent the relationship between uncertainty and risk in the process of LNG ship loading and unloading,and the strength of their mutual influence.