Motion Distribution Analysis And Extreme Prediction Of Semi-submersible Platform Based On Prototype Monitoring Data

Semi-submersible platforms are important engineering equipment in offshore oil and gas development.Its good motion performance and structural stability are suitable for oil and gas exploration in the South China Sea and other deep seas.At present,numerical simulation of hydrodynamics and model tests are the main methods for the structural design of offshore platforms.Due to the complex and variability of marine environmental loads,the nonlinear behavior of floating structures,moorings and other structures,and the scaling effect of structures,hydrodynamic mechanical simulation and model testing still cannot guarantee the reliability of the platform design.Prototype monitoring on the service structure can obtain real environmental load and structural response data.The monitoring data can be used to verify the design index of platforms,and provide the safety of the on-site operation.Therefore,prototype monitoring has become an important method to assist the structural analysis and structural safety of offshore platforms.This paper uses a prototype monitoring system established on a semi-submersible platform in the South China Sea to carry out statistical analysis and motion prediction research on the long-term monitoring data.The main content is carried out as follows:1.Based on the long-term monitored six-degree-of-freedom motion response data,the daily motion response extreme value is extracted and the generalized extreme value distribution model is introduced.Then analyze the distribution law of the one-year response data,and verify the distribution type of the measured data through the KS test method.It is found that except for the heave which conforms to the Gumbel distribution,the rest of the motions conform to the Weibull distribution;2.Based on the obtained motion distribution laws,the extreme values with different return periods are predicted.Compared with the design index,it is found that the once-in-decade extreme value of the heave has exceeded the design indicators,which is the platform design and site operation provides the supports;3.Use deep neural network to build the relationship model between marine environment loads and motions of platform wind and wave actors are selected as the main load parameters,and motions as the prediction targets,short-term deep learning prediction models is established,which are respectively applied to different working conditions of the platform It provides a good basis for the safety of platform operations.