Simultaneous local optimization and coordination of dynamical large-scale systems

To deal with the computational difficulties existing in the conventional methods for large-scale dynamic optimization problems, the paper presents a novel dynamic problem solver for hierarchical control of a class of large-scale dynamical systems by means of a Hopfield-like neural network (LHCNN). The LHCNN consisting of upper layer coordination neural network (UCNN) and lower layer subsystem optimization neural networks (LONN) has the feature of inherent ease for realization by an analog integrated circuit and the property of global convergence. Moreover, the UCNN and LONN can work simultaneously to give the optimal controls and optimal states. Therefore, the LHCNN has high efficiency and is more suitable for real-time industrial applications.