提出一个处理非线性不等式约束优化问题的有效可行SQP算法.每一步迭代,只需求解在近似积极约束指标集下的一个二次规划子问题和一个线性方程组,该方法有效的避免了马太效应.在无严格互补假设条件下,证得算法是全局收敛和超线性收敛的.数值试验表明该算法是有效的.
An effective and feasible SQP algorithm was presented for processing the problem of inequality constrained optimization.During every iteration cycle,only a secondary programming subproblem and a system of linear equations were to be solved within a subset of the constraints estimated as active so that the Maratos effect could be avoided effectively.It was proved without the assumption of strict mutual supplement that the proposed algorithm would be of global convergence as well as superlinear convergence.The numerical result showed that this method was effective.