| The oil and gas reserves in the polar region are very considerable,and with the global warming,the navigation time of the northeast and northwest routes increases significantly,which also greatly improves the possibility of exploiting polar oil and gas resources.But in the polar region,the semi-submersible platform for oil and gas exploitation will inevitably collide with the sea ice,which also poses a great threat to the safety of the semi-submersible platform.Therefore,it is very important to study the collision damage between sea ice and semi-submersible offshore platform.In this paper,the collision damage of sea ice and semi-submersible offshore platform is studied based on finite element method.The main content of this paper are as follows:(1)An elastic brittle material model of iceberg sea ice suitable for collision between iceberg and offshore platform is established.The model is compiled according to the failure criterion and the modification of hydrostatic pressure.The material secondary development based on finite element LS-DYNA solver is realized by FORTRAN language.It is embedded in LS-DYNA user-defined material to generate a solver with elastic brittle iceberg sea ice material.The feasibility of the self-defined material model is verified.(2)The finite element model of semi-submersible offshore platform is established,and the collision simulation of sea ice and platform is carried out by using the established elastic brittle iceberg sea ice material model,and the collision damage of platform is analyzed.Considering the mass,velocity and thickness of the iceberg,the influence of the three factors on the damage of the platform is studied.(3)Using the platform damage data of finite element collision simulation,the machine learning method is used to predict the platform damage,and the preliminary prediction results of different algorithms are analyzed.Then the random forest regression algorithm is selected as the main prediction algorithm to adjust the parameters in detail,predict the platform damage,and analyze the prediction effect. |