摘要
本文以增广拉氏函数为基础,采用序列二次规划和信赖域策略,提出了一种新的拟牛顿乘子法,克服了Powell型约束变尺度法的数值不稳性和Maratos效应。文中给出了其算法程序QNM01的数值实验结果与应用实例。
In view of the drawbacks in the variable metric method for constrained optimization, a new algorithm, the quasi-Newton multipliers method based on the augmented multiplier method, is proposed. A different strategy for sequential quadratic programming is adopted and an algorithm program QNM01 is developed. Numerical results and an example in engineering optimization design are given.
出处
《华中理工大学学报》
CSCD
北大核心
1990年第5期105-110,共6页
Journal of Huazhong University of Science and Technology
关键词
机械设计
拟牛顿乘子法
二次规划
Constrained optimization
Quasi-Newton
multiplier method
Quadratic programming
Engineering optimization design
