The Design of Model-Based Hinf Compensator and its Application on Ships

• Main Authors: Luo, Chung-ching, 羅章才
• Other Authors: C.N. Hwung
• Format: Others
• Language: zh-TW
• Published: 1998
• Online Access: http://ndltd.ncl.edu.tw/handle/96720226534363355787

碩士 === 國立成功大學 === 造船工程學系 === 86 === In this paper, we realize H( control theory with Model-Based method to apply more easily. Inthe concept of the weighting matrices, we offer a way in series from the concept of loop shapin, and regulate the sensitivity matrices, the complementary sensitivity matrices, and the controenergy matrices to achieve the desired performance of the closed-loop system. We can constrict the H( norm from the system input to the controlled system output based on the Model-based H(controller we offer. So we can get the system with both rejecting the disturbance and achieving the desired performance. Because of H( control is powerful to the system uncertainty (with disturbance), but the desired performance of the controlled system can't be certainly achieved, we try to develop H(/LTR from LQG/LTR theory. However the target feedback loop has been developed two kinds of design method:（一）Loop transfer recovery at the plant input and（二）Loop transfer recovery at the plant output, and compare the difference above. Loop shaping and H(/LTR is the same originally to reject the disturbance and achieve the desired performance. Two kinds of the target feedback loop we offer in the paper is discussed the system response of the noise at the plant output and the disturbance at the plant input, and we discuss loop transfer recovery in the method of two designed gain matrices with H( controller. The control gain matrices K and observer gain matrices H can be solved by two Riccati equations. In the paper, we offer the design method in series of selecting the weighting matrices to ensure the target feedback loop will be recovered. So the closed-loop system can achieve the desired performance and reject the disturbance at the same time. In the paper, we offer two simulations to express the design process of H(/LTR, and ensure the controller can be realized and powerful.