This paper presents a new controller design method for learning a nonlinear function existing between real plant and controller outputs in the practical dynamic systems. A new controller is based on the FMFNN(Fuzzy Membership Function based Neural Net...
This paper presents a new controller design method for learning a nonlinear function existing between real plant and controller outputs in the practical dynamic systems. A new controller is based on the FMFNN(Fuzzy Membership Function based Neural Network) being able to learn nonlinear functions well. FMFNN architecture can a) enhance the capability of learning, b) obtain optimal control rules modified by learning. The proposed controller based on the FMFNN is applied to 3rd order linear system and non-linear system. Simulation results show the improved performance such as the stability, response speed, reduced tracking errors.