Novel synchronization of discrete-time chaotic systems using neural network observer.

This paper presents a new approach to solve the synchronization problem of a large class of discrete chaotic systems. The chaotic systems can be reformulated as an appropriate class of linear parameter varying (LPV) systems. The synchronization problem for this class of nonlinear systems is revisited from a control perspective and it is argued that the problem can be viewed as an observer design problem. Then, based on the LPV representation, a neural network observer-based approach is proposed to solve the synchronization problem. The simulation results show the advantages of combining the LPV techniques and the neural networks to determine the appropriate observer gain within the context of chaotic system synchronization.