Study of Identification and Position Control of AC Servo Motor by Using Neural Network

• Main Authors: Din-Sam Wei, 韋丁山
• Other Authors: 洪瑞鴻
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/43424335273652353042

碩士 === 國立臺灣海洋大學 === 機械與輪機工程學系 === 92 === The design and analysis of the traditional control theory needs the description of the mathematics model of each interrelated parameter of the system. But these parameters usually limited in certain circumstances, and uses assumption method to obtain its mathematics model. Since it is difficult to obtain the exact mathematics model during the actual nonlinear environment, it leads to low practicality. Neural Networks have the abilities to learn, to be tolerant of error and to approximate nonlinear function, therefore it is specially suitable for the identification of the nonlinear system. In this thesis, we have designed a position control system of AC servo motor by using Visual Basic, and set up motor’s model using back-propagation algorithm with momentum. The modeling results of input-output’s pair are stored into the computer, to regular the control voltage of motor in order to obtain the best position response. From the experiment, we can obtain good position control response from the modeling results as the judgments rule to regulate the voltage without any additional design control rule.