一个不等式约束问题的SQP方法及其收敛性  

A SQP Method for Inequality Constrained Optimization and Its Convergence

作　　者：解才先[1] 朱宁[1] 

机构地区：[1]桂林电子科技大学数学与计算科学学院,广西桂林541004

出　　处：《汕头大学学报（自然科学版）》2010年第1期17-23,共7页Journal of Shantou University：Natural Science Edition

摘　　要：提出一个关于不等式约束问题的SQP算法,其效益函数为非可微精确罚函数,罚因子具有自动调节性.通过求解一辅助线性方程组,获得二阶修正步,并利用弧式搜索,建立了问题的一个可行下降算法.在一定的假设条件下,证明了算法是全局收敛的,并且具有超线性收敛速度.In this paper, a SQP method, in which the merit Function is nondifferentiable exact penahy function, is presented to solve inequality constraint. The penalty is adjusted automalically. The arc-search and the second-order correction, which is obtained by solving an auxiliary linear equation system, are used Io oblain a feasible descent algorithm. Under some suitable assumplions, it is proved thai the convergence of the algorithm is global as well as superlinear.

关 键 词：SQP方法 弧式搜索 非可微精确罚函数 全局收敛性 超线性收敛性 

分 类 号：O221.2[理学—运筹学与控制论]