| Summary: | 碩士 === 國立臺南大學 === 機電系統工程研究所碩士班 === 100 === However, after NCD system identification is finished, there are still nonlinear terms that can not be identified, such as friction, external noise, etc. If controller design is done in such condition, defects will exist. Therefore, this study proposes a self-tuning PID controller whose parameters can be tuned timely. This self-tuning PID controller is mainly divided into two artificial neural networks to achieve design of self-tuning PID controller. The first artificial neural network conducts system identification for the controlled system by use of RBFNN (Radial Basis Function Neural Networks), to which the most important point is to obtain the Jacobian sensitivity signal of the control system. And the second artificial neural network tunes the KP, KI and KD control parameters of the PID controller by use of the sensitivity signal of the control system obtained by the RBFNN.
Finally, taking the induction motor V/F speed control as platform, apply the structure in this study to the velocity loop. Validate its feasibility by MATLAB simulation. At last, by actual operation, observe its response adding external changed loads when the motor speed achieves a steady state. The results proves that the intelligent PID control design proposed in this study has a better robustness than the traditional PID control design.
|
|---|
