| Summary: | 碩士 === 大同大學 === 電機工程學系(所) === 99 === A PID control scheme via combinations of fuzzy neural network (FNN) and self adaptive differential evolution algorithm with chaos theory (SADEC-PID) is proposed in this thesis. The SADEC-PID controller is consisted of a FNN estimator, a PID controller and a SADEC optimizer. The PID controller is the main effort which uses the error, integral of the error, and derivation of the error with the corresponding parameters to control uncertain nonlinear systems. The FNN estimator is utilized as the tuning-tool for each parameter of the PID controller. The SADEC optimizer is used to optimally select parameters of the FNN estimator. In order to improve the convergence of SADEC optimizer, firstly, a chaotic sequence based on logistic map is introduced to adjust mutation factor. Next, a dynamic updating of population strategy is utilized in process of evolution that responses more flexibly to best value of the current. Furthermore, a self-adaptive crossover probability factor is presented to improve the diversity of population and the ability of escaping from the local optimum. Finally, according to the index of mean-square-error (MSE) to determine the SADEC optimizer whether it will be continuous evolution or correction. The results of the simulation are implemented to verify the effectiveness of the proposed controller.
|
|---|
