Chaotifying Unknown Dynamical Systems via Feedback Control Based on Neural Networks

摘要

In this paper, we study the problem of making a nonchaotic dynamical system chaotic when the system model is unknown. We propose that the unknown system can be identified by using neural networks (e.g. radial basis function neural networks), and then based on the identified model, a state-feedback controller can be designed to make all the Lyapunov exponents of the controlled system strictly positive. The designed controller can drive the unknown system chaotic. Simulations demonstrate the effectiveness of our algorithm.