一类非线性大系统递阶优化的神经网络方法

Hierarchical Optimization of Nonlinear Large-Scale Systems:A Neural Network Method

本文给出了一种求解一类非线性大系统递阶优化问题的神经网络模型，克服了非线性大系统优化中的对偶间隙与不可分性问题，并且该神经网络具有全集成化的特点，易于硬件实现，其协调网络和局部优化网络同步工作，具有很高的求解效率，适宜于系统实时优化应用．

A neural network for optimization of steady-state large-scale systems is devised in the paper. The proposed method overcomes the duality gap and the non-separability difficulties existing in the nonlinear large-scale optimization problems. Furthermore, its coordination network and local optimization networks work simultaneously to give the optimal solutions to the original problems, so it has high efficiency and is suitable for real-time applications.