可探测问题不可行性的无滤子逐步二次规划方法  

A FILTER-FREE SEQUENTIAL QUADRATIC PROGRAMMING ALGORITHM WITH INFEASIBILITY DETECTION

作　　者：刘东梅 刘新为[1] 阎爱玲[1] 

机构地区：[1]河北工业大学理学院,天津300401

出　　处：《高等学校计算数学学报》2017年第3期212-231,共20页Numerical Mathematics A Journal of Chinese Universities

基　　金：国家自然科学基金(11271107;11671116);河北省自然科学基金(A2015202365)资助项目

摘　　要：1引言 本文考虑如下非线性约束优化问题 min f（x） s,t．c（z）≤0，A filter-free sequential quadratic programming （SQP） method is pre- sented for nonlinear inequality constrained optimization. The method computes a search direction by solving subproblems based on an exact penalty function and has the important feature of infeasibility detection when it is employed to solve infea- sible instances. Furthermore, in each iteration, the step is selected such that either the value of objective function or the measure of constraint violations is sufficiently reduced. A nonmonotone technique originated from the solution of unconstrained optimization is applied to accelerate the algorithm. Under standard assumptions, global convergence of the proposed algorithm is established. The preliminary num- erical results are also presented to show the efficiency of the proposed algorithm.

关 键 词：逐步二次规划 非线性约束优化问题 滤子 探测 

分 类 号：O221.2[理学—运筹学与控制论]