非线性等式优化的一种非单调SQP滤子算法

摘要

SQP滤子方法是解非线性规划的一种较为有效的方法,但是滤子方法也会遇到Maratos效应.采用非单调技术来避免Maratos效应,并采用降维的Byrd和Omojokun方法来计算试探步.在一定条件下,给出了全局收敛性证明,数值试验表明该算法有效.%Nonlinear constrained optimization was efficient and robust solved by the SQP filter approach. But,the so-called Maratos effect was suffered. A non-monotone trust region method was presented. The step was computed by the Byrd and Omojokun scheme. Global convergence was proved under certain conditions.