Decentralized Neuro-Fuzzy Control of a Class of Nonlinear Systems
A decentralized control based on recurrent neuro-fuzzy networks is proposed for a class of nonlinear systems. It consists of an adaptive component and a uncertainty compensation component. First the control law is designed using the state feedback, and the semiglobal stability is established. Then, by means of a highgain observer, this control law uses only the output feedback. The main features of the proposed scheme are its robustness against uncertainties and its simplicity of implementation. To illustrate the proposed scheme, experiments on a 2-degree-of-freedom robot are included.
KeywordsIEEE Transaction Tracking Error Output Feedback Output Feedback Control Decentralize Control
Unable to display preview. Download preview PDF.
- 10.T. Takagi and M. Sugeno (1984) Fuzzy identification of systems and its applications to mod-eling and control. IEEE Transactions Systems, Man, and Cybernatics, SMC-15: 116-132.Google Scholar
- 11.L. Wang (1994) Adaptive fuzzy systems and control, design and stability analysis. Prentice Hall, Uppr Saddle Rivier NJ.Google Scholar
- 25.K.I. Funahashi and Y. Nakamura (1993) Approximation of dynamical systems by continouos time recurrent neural networks. Neural Networks. 801-806.Google Scholar
- 26.G.A. Rovithakis (2000) Adaptive control with recurrent high-order neural networks: theory and industrial applications. Springer.Google Scholar