A Hybrid method of NN and Fuzzy Theory for a PID controller of LEO Satellite Tracking Systems

• Main Authors: Cheng-Hung Hung, 洪誠鴻
• Other Authors: Te-Jen Su
• Format: Others
• Language: zh-TW
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/ksbeew

碩士 === 國立高雄應用科技大學 === 電子工程系碩士班 === 102 === This thesis presents a hybrid method of Neural Network (NN) and Fuzzy theory for a Proportional-Integral-Derivative (PID) controller of Low Earth Orbit (LEO) satellite tracking systems. For traditional PID controller, parameter adjustment depends on experience and the present situation which is very inconvenient. Neural network has good capability of adaptive learning and prediction for various applications in controller design and system identification. NN with learning ability that applied to PID controller in this thesis can help to search the optim al parameters kp、ki and kd in real time that overcomes the shortcomings in traditional PID controller. Neural networks establish models with present resources that lead to lack of global accuracy which may cause considerable errors and not be practical. The fuzzy theory is adopted to adjust the rotation speed of motors to obtain accurate outputs. Finally, we use resolver feedback mechanism of actual angle indicator as a basis for judgment performance advantages and disadvantages. Experimental results demonstrate that the proposed NN-Fuzzy PID controller results in raising control precision efficiently for position, speed and tracking control.