Surrogate-Assisted Evolution Algorithm Based on Label Propagation Guidance and Region Adaptive Integration
Agent-assisted evolutionary algorithms have been widely used to solve computationally expensive optimization problems.In order to improve the ability and efficiency of the pre-screening solution of the agent model under the condition of limited computing resources,the paper proposes an agent-assisted evolution algorithm based on label propagation guidance and regional adaptive integration,which is divided into two stages:global search and local search.In the global search stage,the multiple screening criteria(MSC)is proposed.Two proxy models are used to predict the fitness values of the classified populations,and the populations are screened again according to the best fitness of the parents and offspring.The local search phase uses the SMOTE(synthetic minority over-sampling technique)method to generate dynamic local populations and test samples,select individuals for evaluation based on adaptive integration of the performance of the two proxy models in the most promising areas.Finally,the proposed method is compared with other advanced agent assisted evolution algorithms in 8 test problems and airfoil design optimization problems,showing that the proposed method has better convergence.
万方数据
