On the suboptimal solution for large scale non-linear control systems using neural networks

Abstract

This work is aimed at looking into the determination of optimal control strategies of large scale non-linear systems. The basic idea consists of the decomposition of the global system into interconnected subsystems. We consider non-linear systems whose subsystems are linear with respect to their state and control; in other words, non-linearities are located in the interconnection terms. Then, for each subsystem, local optimal feedback gains are easily expressed in terms of interconnection vectors. Such optimisation has been successfully achieved using a neural net yielding the optimal gain at each step time. Some numerical examples are considered in order to illustrate the proposed approach.