非线性最优化问题的复合形旋转方向搜索法

针对不等式约束非线性最优化问题,分析了传统复合形算法收敛速度慢的原因,提出了一种称为复合形旋转方向搜索的新算法,给出了算法的迭代计算流程和程序框图.该算法与传统复合形法的主要区别在于:①迭代计算不以中心点作为复合形收敛中心,而以最好点作为复合形收敛中心;②迭代计算的映射点不在最坏点与中心点的连线方向选取,而在最好点与复合形各顶点的连线方向选取.

A Revolving Direction Search Algorithm of the Complex Shape on the Nonlinear Optimization Problems

Aiming at the nonlinear optimization problems with inequality restriction, we analyzed why the convergence speed of the traditional complex shape arithmetic was slow. A new algorithm referred to as the revolve direction search algorithm of the complex shape is put forward here and its iterativing computation flow and procedure chart are presented. The chief differences between the new algorithm and the old of the traditional complex shape are. firstly, the convergence center of the complex shape is not at the central point but at the best point searched out during the process of iterative computation; secondly, the mapping point is not selected in the direction of the line linking the worst point with the center of the complex shape, but in the direction of the line linking the best point with each acme of the complex shape.