Explicit topology optimization of novel polyline-based core sandwich structures using surrogate-assisted evolutionary algorithm

Inspired by venation patterns of leaves, the new polyline-based core sandwich structures (PCSSs) are developed in this paper. A novel approach is proposed under explicit topology optimization framework based on surrogate-assisted evolutionary algorithm. Specifically, a multi-component topology description function is defined to characterize the shapes of polyline-based cores, and therefore the dimension of design variables in topology optimization of PCSSs is reduced to only tens of dimensions. To solve the optimization problem, a feedback mechanism-driven surrogate-assisted differential evolution algorithm is proposed, where two search phases, i.e., global and local search phases, are respectively implemented in sequence to achieve a good balance between global exploration and local exploitation. Specifically, the global search phase is utilized to not only explore unknown areas for locating the promising regions effectively, but also improve the accuracies of surrogates in specific promising areas. The local search phase is used to intensively search the local promising regions for accelerating the convergence rate. A feedback mechanism is proposed to drive the adaptive search in the global search phase based on three different kinds of feedback information. Based on the superiorities of constraint satisfactions of PCSSs, a volume domination-based epsilon-constraint handling method is employed to handle volume and other constraints differently in an appropriate way. Numerical examples are provided to demonstrate the effectiveness and efficiency of the proposed method. (C) 2020 Elsevier B.V. All rights reserved.