用于约束多目标优化问题的热力学遗传算法

Thermodynamics genetic algorithm for constrained multi-objective optimization

为了保持所求得的约束多目标优化问题Pareto最优解的适应度与多样性,在NSGA-II基础上提出了一种用于求解有约束的多目标优化问题的热力学遗传算法。结合热力学中自由能与熵的概念,利用热力学中熵与能量的竞争来保持种群的适应度与多样性的平衡,设计了热力学算子。根据非支配排序Pareto分层结构建立分层小生境来改进选择算子,弥补了选择算子不足。实验结果表明:该算法不仅得到的解在空间分布均匀,收敛性好,同时解集具有较广的分布空间。

In order to maintain the fitness and diversity of the solution of the constraints multi-objective optimiza- tion, This paper proposes the solving constrained multi-objective thermodynamics genetic algorithm based on the NSGA-II. Binding free energy and entropy in thermodynamics, using entropy in thermodynamic and energy compe- tition to maintain the balance of population fitness and diversity, it designs thermodynamics operator. The layered niche technology is used to improve the selection operator. The numerical examples show that the proposed approach is provide good performance in terms of uniformity and broader of solutions.