一类可行的两阶段SQP方法  

A Feasible Two-phase SQP Method

作　　者：刘美玲[1] 

机构地区：[1]上海电机学院数理教学部,上海201306

出　　处：《数学理论与应用》2015年第1期71-82,共12页Mathematical Theory and Applications

基　　金：上海高校青年教师培养资助计划项目(ZZSDJ13008)

摘　　要：提出了解约束优化问题的一类可行的两阶段SQP滤子算法.利用一类两阶段序列二次规划方法计算试探步,而用滤子接受准则选择接受试探步.针对二次规划子问题的不可行问题,对其约束引进参数进行了可行化处理,可以省略可行恢复项,节省了计算时间.在一般条件下,算法具有全局收敛性.最后,数值试验显示了较好的结果.A feasible SQP algorithm combined with the filter method is proposed for constraint optimization problems.This iterative mechanism is based on a two- phases SQP method. Introducing filter method is used as an acceptable criterion for the trial step. For the infeasibility of QP subproblem,the introduction of some parameters makes the problem always feasible,so a feasibility restoration phase is not needed any more. This can save run time in implementation of the algorithm. The algorithm converges globally under mild assumptions and its efficiency is showed by some numerical tests.

关 键 词：非线性规划 可行性 滤子 收敛性 

分 类 号：O224[理学—运筹学与控制论]