摘要
本文讨论了一类非凸稳态大系统的递阶优化方法,采用增加部分约束罚项的思想,提出一种既能保持原问题的可分性结构,又能将原问题凸化的构造增广拉格朗日函数的新方法,证明了凸化后的新问题与原问题之间的等价性关系,研究了它们的递阶优化算法,证明了所给算法的局部收敛性,讨论了算法所具有的收敛速度。
Using increasing partial contrained penalty terms, the paper offers anew hierarchical optimization method for a Kind of nonconvex steady-statelarge-scale systems. It, basing on new augmented Lagrangian function, canKeep the separable structure of original problems and alse convexify theoriginal problems. The equivalent relation between convexifying problems andthe original is proved. The paper also studies the hierarchical optimizationalgorithm and the result shows it is locally convergent and also estimates itsconvergence rate.
关键词
大系统
递阶优化
凸化技术
Large-Scale systems
hierarchical optimization
convexifying techniqe
lagrangian function
