Diverse competitive design for topology optimization

Abstract

The traditional topology optimization method aims at finding the optimum design of material distribution. This paper explores the problem formulation and solving approaches to find multiple Diverse Competitive designs for Topology Optimization problem (DCTO). Diversity constraints are used to guarantee the difference between multiple designs. Three graphic diversity measures are presented to set the desired diversity. Four different topology optimization problems including the compliance minimization, the compliant mechanism problem, and the heat conduction problem are used to demonstrate the effectiveness of DCTO. We compare three types of DCTO algorithms for generating four designs and conclude the recommended algorithms. The ranges and effect of the diversity measures are discussed for helping select a reasonable diversity constraint to balance the performance and diversity. Threshold projection and length scale control are also available in the DCTO approach to get a binary solution and avoid small members and holes.