基于改进RBF代理模型的海上风机单桩基础优化设计

Optimal design of offshore wind turbine monopile foundation based on improved RBF surrogate model

文章提出了一种基于径向基函数(RBF)神经网络结合粒子群算法(PSO)的海上风机单桩基础优化设计方法。该方法考虑局部冲刷的影响,提高了优化计算效率和结构安全性。首先,基于6.45 MW海上风机单桩基础工程实际问题,以泥面以下部分的桩径、壁厚和桩长为设计变量,采用拉丁超立方抽样法选取样本点;然后,建立考虑局部冲刷的单桩基础有限元模型并计算响应值,构建RBF代理模型,结合PSO算法进行全局寻优。寻优结果表明:考虑局部冲刷影响的海上风机单桩基础优化设计结果更趋于安全;适当增加桩径和壁厚比增加桩长更加经济;使用RBF代理模型能显著提高优化效率。

To improve the optimization efficiency as well as structural safety,a hybrid structural optimization method for design of offshore wind turbine monopile foundation is proposed by combining radial basis function(RBF)neural network and particle swarm algorithm(PSO).Taking the monopile foundation of a 6.45 MW offshore wind turbine as a typical example.Firstly,the pile diameter,wall thickness and pile length below the mud surface are used as the design variables,and the sample points are selected by the Latin hypercube sampling method.Then,the finite element model of the monopile foundation considering local scour is established and the response value is obtained,so as to construct the RBF surrogate model.Finally,the PSO algorithm is introduced to optimize the design variables.The results show that the influence of local scour needs to be considered in the optimization design of monopile foundation,and at the same time,it is more economical to increase the pile diameter and wall thickness than to increase the pile length.In addition,the surrogate model shows higher efficiency than using detailed finite element calculations.