An integrated design framework of floating wind turbine based on surrogate-assisted many-objective optimization

The optimal design of floating wind turbines (FWTs) is an integrated many-objective task in which the costs of components, power quality, and other system performance are the primary concerns. The application of multi-objective evolutionary algorithms (MOEAs) is challenging due to the function evaluations are computationally expensive. Therefore, we propose a surrogate-assisted constrained many-objective optimization framework (SCM) to enhance the efficiency of MOEAs in solving the problem. Firstly, an initial surrogate model is constructed using data generated by the MOEAs. Then, a cooperative surrogate-assisted evolutionary search strategy is proposed. It uses the surrogate model to filter out poor solutions and provide more evolutionary resources for solutions with better convergence, diversity, and feasibility. Further, a balanced sampling strategy is proposed to update the surrogate model online. Moreover, when the population evolves into the later stage, the MOEAs are used instead of cooperative surrogate-assisted evolutionary search to ensure that the solutions are uniformly distributed on the true Pareto front. Finally, the SCM is verified in an integrated many-objective optimal design (IMOOD) problem. Under the critical design load case (DLC), SCM can obtain a FWT design set with convergence, diversity, and feasibility in a smaller number of function evaluations (FEs).