Font Size: a A A

Research On Indirect Evaluation Method Of The Ship Oil Spill Damage Compensation

Posted on:2013-01-11Degree:MasterType:Thesis
Country:ChinaCandidate:W C ZhuFull Text:PDF
GTID:2231330362975293Subject:Marine Engineering
Abstract/Summary:PDF Full Text Request
As one of the key technology of the oil spill control and treatment, damage assesssment andcompensation connects directly with cleanning efficiency and environmental restoration. In viewof the limitation of the the direct evaluation method in practical application, this study used PSO-SVM model to simulate the nonlinear relationship between the compensation and influence factorsamong history cases. It provides a indirect and effective evaluation method for the oil spill damagecompensation.Firstly this study analyzed the factors completely, concluding the diagram of analytichiberarchy based on the formers. At the same time, this study designed a questionnaire to obtainthe experts’ experience. Then the factors’ weights were got by the Analytic Hierarchy Process(AHP)method.Secondly the article put forward five dominating factors such as oil amout、oil specificgravity、sea situation、oil location、oil clean up situation to assess the damage. By analyzing thecompensation cases of IOPC Fund, five primary factors and compensation were treated with fuzzymathematics, which were extracted as the input and outout vector of model.So it was a good workfor the next subsequent model.Finally the particle swarm optimization (PSO) was introduced into support vector machine(SVM) model to optimize its parameters about c and g. Using the PSO-SVM model to forecast testset, the result shows that the relative errors rate are under10%. And comparing the result withthose by traditional SVM and GS-SVM, PSO-SVM model also gets a better effectiveness andsuperiority. So PSO-SVM model has a reference practical value for the oil damage assesssment.
Keywords/Search Tags:oil spill, damage compensation, indirect assesssment, fuzzy mathematics, particle swarm optimization, support vector machine
PDF Full Text Request
Related items