Risk Assessment Of Inland Waterborne Transportation Safety Using Improved Bayesian Network Model Based On Data Mining

China has completed the construction of a relatively sophisticated inland water transportation system,which has achieved an unprecedented growth in the past few decades.However,inland waterborne transportation accidents occur frequently in China with serious consequences.The Orient Star shipwreck was a catastrophic event with huge casualties and property losses and aroused widespread concern about inland water traffic safety in China.At 21:30 on June 1,2015,the "Oriental Star" passenger ship belonging to Chongqing Oriental Steamship Company encountered a rare strong convective weather on the way from Nanjing to Chongqing,and sank in the Jianli waters of Hubei Province in the middle reaches of the Yangtze River.The serious casualties and huge property losses caused by the " Oriental Star" shipwreck accident once again highlight the importance of the safety of China’s inland waterborne transportation.The safety of China’s inland waterborne transportation needs to be resolved.The purpose of this thesis is to explore the causes of the risk of China’s inland waterborne transportation,formulate phase strategies,reduce the occurrence of such risk accidents,improve the safety of inland waterborne transportation,and finally contribute to the development of China’s inland waterborne transportation.The main work of this thesis includes the following four aspects:Based on the accident analysis reports of the China’s inland waterborne transportation collected for nearly 30 years,the text database is built.After data preprocessed,plain text data is obtained.Through text mining,the risk factors are identified and extracted from the accident analysis report,and the word cloud is shown visually at the same time.According to the relevant attributes of risk factors of at the China’s inland waterborne transportation,they are divided into four categories: ship factor,environmental factor,human factor and accident factor.Each category is subdivided into 2～5 Sub-variables,and determine the threshold of each indicator based on the historical research.On this basis,the association mining based on FP-Growth algorithm is used to explore the relationship between risk variables and draw the associated network graph.The Bayesian network model is constructed based on the risk variable association network obtained by association rule mining to do mutual information analysis,Kullback-Leibler analysis,most likely interpretation and sensitivity analysis.According to the above analysis results,combined with the connection between the various nodes,the degree of the risk factors influence on the consequences of the accident is judged.The results reveal that overloading or improper loading,poor navigation visibility,poor sailor quality,and insufficient government supervision of shipowners and companies are all key factors of accidents.Finally,incorporating the actual circumstances of domestic inland waterborne transportation operations,the study puts forward targeted suggestions to the government and relevant departments.