Uncertainty Analysis And Optimization Design Of Offshore Wind Turbine Power And Support Structure

Driven by the goal of "peak carbon dioxide emissions" and "carbon neutral",wind power has become one of the key points of energy transition.For offshore wind turbines,continue to go deep sea is the trend of the future development of the industry,while the continuous development of larger turbines is also an urgent need to solve the technical problems.Large-scale also means that the system reliability and structural reliability of the wind turbine is facing great challenges.Large wind turbine wind wheel diameter and tower height can reach more than hundred meters.For this large equipment,there are inevitable errors in the actual manufacturing and assembly process,it also faces complex and uncertain service en vironment.All of errors will lead to fluctuations in the overall system or structure of the wind turbine and affect the reliability and safety of wind turbine.It is necessary to analyze the uncertainty of wind turbine structure and system,accurately mea sure the impact caused by uncertainty factors for the wind turbine design stage.Based on the study of NREL5 MW offshore wind turbine,this paper introduces uncertainty analysis to measure the impact of uncertainty parameters on the output power and the re sponse of the support structure,optimizes the reliability design of the support structure.The main research content is divided into the following three parts.(1)Uncertainty propagation analysis of offshore wind turbine power.Firstly,the aerodynamic characteristics of the wind turbine are comprehensively analyzed,and the transformation relationship between the rotor energy and torque power of the wind turbine is derived.Then the NREL 5MW offshore wind turbine model is built in the wind turbine simulation software,and the power simulation is carried out under normal power generation conditions.In order to obtained the uncertainty of wind turbine power generation,the uncertainty variables of external environment such as sea state and wind condition are c onsidered,which are all measured by probability models.To obtain the uncertainty distribution of the power,manifold learning approach is adopted for the analysis.First,getting sample points in the parameter space which are extracted by optimal Latin h ypercube sampling,then the power is obtained by wind turbine simulation software,the cumulative distribution function curve is obtained after discretization.After that,we can get the eigenvalues and eigenvectors through manifold learning dimensionality reduction,and then the mapping relationship between uncertainty variables and eigenvalues is established by surrogate model,so that the power can be efficiently obtained by Monte Carlo simulation.The probability box model of power is obtained efficient ly through Monte Carlo simulation,and the uncertainty propagation of power is completed.(2)Uncertainty propagation analysis of wind turbine support structure based on dimensionality reduction decomposition.Firstly,a finite element model of the offshore wind turbine’s monopile support structure is constructed and verified to be valid,which is used as the system model for uncertainty analysis.Uncertainty variables such as structural dimensions,external environment,physical properties,material properties and soil properties were taken into account.Carrying out the modal analysis and static analysis,obtaine responses as the system output such as natural frequency,stress and deflection.The statistical moment information of the subsystem is obtained by calculating the system response at each configuration point through the reduced-dimensional integration method,so as to combine the statistical moment information of the original system and complete the uncertainty propagation.Then the accuracy of the propagation results is verified by Monte Carlo simulation.(3)Reliability-based design optimization of wind turbine support structure based on surrogate model.The reliability-based design optimization is mainly based on the finite element model of the support structure built out.Since the finite element calculation is very time-consuming,the optimization is not carried out by the method of real-time update through finite elements.Getting sample points through optimal Latin hypercube sampling,the calculation of finite analysis is obtained,and then establishe the mapping relation between the sample points and the system output by Kriging surrogate model.The Kriging surrogate model is transformed into the objective function,limit state equation and constraints in the reliability-based design optimization.Then the size of the support structure is optimized through the SORA decoupling method based the guaranteed reliability.Finally,the optimized finite element model is subjected to modal analyses,static analyses and flexural analyses to verify the reliability of the optimized results.