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.
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
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
SQP method
are-search
nondifferenliable exact penally function
global convergence
superlinear convergence
