An Observer-Based Neural Network Controller for Chaotic Lorenz System

The paper proposes an alternative control design for the chaotic Lorenz system based on neural networks. The controller is a feedforward neural network trained by a model reference technique. Implementation of the control design requires system states for feedback, while in most of practical applications only the system Output is available. To overcome this problem, a nonlinear observer is used to estimate the states of the system. Simulation results have illustrated the feasibility and effectiveness of the proposed observer-based neural network controller.