基于拟态物理学的约束多目标共轭梯度混合算法

在拟态物理学优化算法APO的基础上,将一种基于序值的无约束多目标算法RMOAPO的思想引入到约束多目标优化领域中.提出一种基于拟态物理学的约束多目标共轭梯度混合算法CGRMOAPA.算法采取外点罚函数法作为约束问题处理技术,并借鉴聚集函数法的思想,将约束多目标优化问题转化为单目标无约束优化问题,最终利用共轭梯度法进行求解.通过与CRMOAPO、MOGA、NSGA-II的实验对比,表明了算法CGRMOAPA具有较好的分布性能,也为约束多目标优化问题的求解提供了一种新的思路.

Constrained Multi-objective Conjugate Gradient Hybrid Algorithm Based on Mimicry Physics

On the basis of mimicry physics APO optimization algorithm,the thoughts of unconstrained multi-objective algorithm MOAPO based on a sequence was adopted into the constrained multi-objective optimization field.The paper proposed a constrained multi-objective conjugate gradient hybrid algorithm CGRMOAPA based on mimicry physics.The outer point penalty function method was adopted as a processing technology of the constrainted problem,and the thought of a aggregation function method was used in to the algorithm,then the constrained multi-objective optimization problem was transformed into the single objective unconstrained optimization problem,finally conjugate gradient method was used to solve the final problem.Through experiment contrasts with CRMOAPO、 MOGA、 NSGA-II,It shows that the algorithm CGRMOAPA has better performance of distribution.Also it provides a new train of thought for the constrained multi-objective optimization problem.