Font Size: a A A

Analysis Of Vessel Collision Results Based On Cluster Analysis

Posted on:2011-10-19Degree:MasterType:Thesis
Country:ChinaCandidate:J J YuFull Text:PDF
GTID:2132360302999369Subject:Traffic Information Engineering & Control
Abstract/Summary:PDF Full Text Request
The collision accidents in maritime traffic occur frequently, which cause not only huge losses on lives and economic, but also the damage of ship structure, the cargo leakage, the pollution of ocean, the loss of life and person injuried and so on. Therefore, it has great significance to analyze occurred ship collisions, summarize the meaningful analysis model from the consequences of ship collision and give reasonable advice to the lead officers to avoid collision or bring down the loss of collision both for security, economy and environmental protection.The paper obtains the detail data of ship collisions based on the translation and collection of the 254 collision accidents reports from home and abroad. The factors may affect the collision are analyzed by stepwise regression analysis, and the factors are clustered to determine the function of collision factors and collision result.The data obtained from the ship collision is treated to be centralization and standardization to eliminate the influence from the order and the dimension. The tonnage, the collision angle, the speed before hitting and the position of collision are determined to be the key factors of the loss of the ship after collisions by stepwise regression analysis. Furthermore, influence coefficient of tonnage and collision angle is obviously bigger than the influence coefficient speed before hitting and position of collision. The tonnage plays a key role in the result of collision loss through separate analysis to different factors.The original data was divided into 13 categories with the mark of the number of weighted loss of the collision effect to tonnage with clustering algorithm. Then, the inverse function model between the loss of ship collision and tonnage is obtained through the non-linearrelation curtained by correlation analysis, nonlinear analysis and curve fitting with the divided data instead of the original sample data.The loss of collision will be littler with the increment of tonnage from the established model. However, the analysis of the collision factor is not very complete due to the limited collision accidents reports. There is great value for further research if the collision accidents reports are rich enough.
Keywords/Search Tags:cluster analysis, collision factor, regression analysis, collision results
PDF Full Text Request
Related items