摘要
本文将滤子法用到SQP方法中,通过减小违反约束度函数值和一个逼近目标函数的函数值来确定试探步是否被滤子接受.此方法不同于其他滤子法,它不需要减小信赖域半径而是通过改变步长因子来保证充分下降性.本文在一定条件下获得了全局收敛性.
In this paper,the filter approach is employed in a SQP algorithm. We reduce the degree of constraint violation and some function to decide whether to accept the trial step,and the function is closely related to the objective function. This algorithm is different from other filter algorithms. The efficient decrease condition is controlled by reducing the step length rather than reducing the trust region radius. Under some conditions, the global convergence property is obtained.
出处
《应用数学》
CSCD
北大核心
2010年第3期670-674,共5页
Mathematica Applicata
基金
国家自然科学基金资助项目(10501009)
广西区自然科学基金(0728206)
广西高校优秀人才资助计划项目
关键词
滤子法
SQP算法
非线性规划
不等式约束
Filter
SQP algorithm
Nonlinear programming
Inequality constrained