HYBRID PID-LIKE SELF-CONSTRUCTING FUZZY NEURAL NETWORK CONTROLLER DESIGN
| Autor: | Ming-hua Cheng, 鄭明樺 |
|---|---|
| Rok vydání: | 2010 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 98 A hybrid proportional-integral-derivative fuzzy neural network (FNNPID) controller is designed for uncertain nonlinear systems in this thesis. The FNNPID controller includes three components which are a PID controller, an FNN estimator, and a robust controller. First, the PID controller is the main controller which uses the error, integral of the error, and derivation of the error with the corresponding parameters to control uncertain nonlinear systems. Next, the FNN estimator is used to estimate the parameters of the PID controller. The structure learning utilizes Mahalanobis distance to construct the structure of the FNN. The parameter learning is applied to adjust the parameters in the FNN via adaptive laws which is proven to be stable by Lyapunov theorem. Finally, the robust controller is proposed to solve the problem of the uncertainties via sliding surface. Different from traditional PID controller, the parameters will be online adjusted in this thesis. The simulations of the inverted pendulum and the second-order chaotic system are compared with that of other controllers to demonstrate the performance of the proposed controller. The experiment result of the inverted pendulum system is implemented to verify the effectiveness of the proposed controller. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|