Nonlinear neural network approach to hierarchical optimization

A specific neural network for hierarchical optimization of a class of nonlinear steady state large scale systems is devised in the paper. The approach proposed eliminates obstacles involved in duality gap and separability which will occure in conventional approach to the optimization problem to be solved, in addition, the neural network is completely integrated and is easy for implementation. As components of the whole neural network, the coordination neural network and local optimization neural networks work simultaneously to solve the problem, which leads to high efficiency and potential applicability for real-time control of the device. Copyright (C) 1998 IFAC.