Many‐objective evolutionary algorithm assisted by a novel angle‐based fitness strategy
Zhuanghua Zhu · Concurrency and Computation Practice and Experience · 2022
Summary Measuring the convergence of solutions is greatly significant for many‐objective optimization problems. Although various methods have been proposed regarding this issue, they still suffer from the performance consistency on different problems. For this issue, this article proposes a novel angle‐based fitness strategy. To be specific, the convergence of solutions is defined with adaptive key solutions, which are determined by the angles between solutions. Based on the novel angle‐based fitness strategy, this article designs a many‐objective evolutionary algorithm assisted by a novel angle‐based fitness strategy. Experiments are conduced to verify the performance of the proposed method in comparison with other state‐of‐the‐art methods. The experimental analyzes illustrate the outstanding performance of the proposed method for many‐objective optimization problems.