Design of controller for nonlinear system using chaotic neural networks
Park Seong-Wook, Bo-Hyeok Seo · 2002
A chaotic neural network is investigated in the context of the control for finite dimensional, nonlinear discrete systems. In the proposed method, two chaotic neural networks control a nonlinear system. Each of the two chaotic neural networks CNN control system is based on a framework of model reference control. Then a closed loop error is observed by the difference of chaotic network output and the model output. The proposed learning method trains the CNNI using the identification error and the closed loop error. Numerical examples are presented for a model system and a hydraulic servo valve system perturbed by Gaussian noise. Those case studies showed that the presented learning method is so useful in practical sense.