一种约束非线性离散优化问题的直接搜索法

本文针时整数梯度搜索方向的局部性以及有时因取整而方向发散的缺点,提出了在离散设计空间中,利用离散坐标邻域点的目标函数值信息和适当的延伸探索直接构造离散搜索方向,并进行离散一维搜索的寻优方法,对此称之为离散直接搜索法(A Discrete Direct Search Method ) ,简称DDSM法。本文运用DDSM法成功地解决了14个工程考题和数学考题,实践表明:DDSM法的求解速度和可靠性比较令人满意,尤其是计葬效率相甘现有的莱些离散优化方法有较明显的提高。

DI SCRETE DIRECT SEARCH METHOD FOR A CONSTRAINED NONLINEAR DISCRETIZATION PROBLEM

Aiming at the limitation in search direction of integer gradient methodand the drawbacks in direction divergence occuring at times in integer treat-ment, this paper has proposed a discrete method with rather good search direction or DDSM, which takes the advantage of the objective function information at coordinate ad jacent points in discrete design space and through proper stretches, and constructs a rather good discrete search direction, With its help we have successfully solved 14 engineering and mathematical probl-ems. Optimization practice has showed that calculating speed and its reliab-ility are more satisfactory, In particular, the calculating efficiency is re-markably raised as compared with present discrete optimization methods,