摘要
提出一种带非线性互补函数的信赖域序列二次规划滤子算法.通过引入滤子概念,避免了罚函数法中罚参数选择的困难.借助非线性互补函数简化了非线性规划问题的KKT条件,并用非线性互补函数代替滤子中的约束违反度函数,在一定的条件下证明了算法具有全局收敛性.数值试验结果表明算法是有效的.
A new method of SQP-Filter with NCP fimction was proposed. The difficulty of choosing the penalty parameter associated with use of penalty functions could be avoided by introducing a new concept of ＂filter＂ . This method simplified the KKT conditions of the problem NLP by using the NCP function, and used the NCP function replace the constraint violation function in filter. Under suitable conditions, the global convergence was proved. The computational results showed that this algorithm was efficient and reliable.
出处
《安徽大学学报（自然科学版）》
CAS
北大核心
2010年第5期16-20,共5页
Journal of Anhui University(Natural Science Edition)
基金
国家自然科学基金资助项目(10861005)
广西壮族自治区自然科学基金资助项目(0991238)
安徽省教育厅自然科学基金资助项目(KJ2010B300)
