基于自适应K阶差分演化的多目标优化算法

Multi-objective Optimization Algorithm Based on Self Adaptive K-order Differential Evolution

差分演化算法已经被成功地用于解决单目标和多目标优化问题,然而它的搜索能力受限于所使用的变异模式和控制参数。为此,提出一种新的变异模式——K阶差分,并引入参数自适应机制,设计了一个基于自适应K阶差分演化的多目标优化算法。与NSGA-II、DEMO的仿真实验比较结果表明,该算法在ZDT测试问题上能获得较好的优化效果。

Differential evolution has been successfully employed to solve single-objective and multi-objective prob- lems. However, its search ability is significantly influenced by its variation patterns and control parameters. There- fore, a new variation pattern named K-order difference is proposed in this paper. Then, a multi-objective optimization algorithm, SKDEMO, which is based on K-order differential evolution and self adaptive parameter adjustment mecha- nism, is designed to solve multi-objective problems. The experiments on five ZDT benchmark problems show that SK- DEMO exhibits better performance than NSGA-II and DEMO.