Study Of Ship Collision Accident Pattern In Estuary Based On Data Mining

Estuary is a joined area of river and sea, and the traffic accident is easier tohappen due to the intense ship traffic and complex navigation environment.Especially collision accident is main form of traffic accident in estuarine waters. Thefatal accidents always cause loss of life, economic loss, environmental damage andother possible unwanted events. The fatal accident is becoming the focus of attentionof people with the development of economy and the improvement of people’s livingstandard, water traffic safety directly affects the country’s economic development andsocial stability. Therefore, carrying out a research on ship collision accident pattern inestuarine waters is of great significance.The paper collects historical data of ship traffic accident based on the accidentinvestigation report, and constructs the mining model of ship collision accidentpattern, and the fuzzy fault tree analysis is used to study the probability distribution ofcollision accident patterns. The specific research work includes the following fouraspects:1. Propelling structured design of accident data according to the characteristics ofship navigation environment in estuarine waters, and building Physical Data Modelon basis of demand analysis, cleaning collision accidents data, then importingprocessing data into Physical Data Model to complete the establishment of thedatabase.2. The different cause factor of collision accident is analyzed according to collisionprocess on basis of the collision accident data statistical analysis, the cause factors ofcollision accident contain human factors, ship, goods, environmental factors andmanagement factors, and the relationship between human fault behavior andinfluence of human fault behavior is deeply studied.3. The Apriori algorithm is introduced to explore ship collision pattern and find outavailable information behind a great quantity of maritime accidents. And accidentpattern mining of entire element, preferential element pattern mining and patternmining based on consequence are carried respectively. The collision pattern whichconsists of collision cause factor and the result of the collision is extracted. 4. The paper builds the Fault Tree of ship collision according to the extractedcollision pattern, and fuzzy fault tree is introduced to estimate the probability of shipcollision. The fuzzy probability of collision is acquired through calculation of eachbasic event fuzzy probability.The ship collision accident pattern with large probability is selected through theanalysis of the fuzzy probability. The specific measures of navigation safetymanagement of estuarine waters can be put forward on basis of the characteristics ofthese collision accident patterns. The qualitative analysis of ship collision accidentpattern research is extended to quantitative calculation, and the study is of theoreticalsignificance.