一种求解非线性约束优化的单变量边缘分布算法

An algorithm for Solving Constrained Optimization Problems with Univariate Marginal Distribution

借鉴罚函数法思想,将建立在Gauss网络的单变量边缘分布算法应用于非线性约束优化问题,提出的新算法突破了传统基于约束保持法或可行规则法的约束处理.且单变量边缘分布是基于搜索空间的宏观层面的进化方法,具备更强的全局搜索能力和更高的收敛率,从而为约束问题的求解提供了一种新的途径.

In view of penalty function＇s idea,application of UMDA based on generic Gassian Networks to the nonlinear constrained optimization problems are described in the paper. The proposed algorithm breaks the classical method of constraint preserving and feasible rule. In addition,EDAs with a stronger global search ability and higher convergence,is macro-level evolutionary approach based on search space. Therefore,a novel way of the constrained optimization problems is proposed.