LQG/LTR, H-infinity and Mu robust controllers design for line of sight stabilization

Download
2015
Baskın, Mehmet
Line of sight stabilization against various disturbances is an essential property of gimbaled vision systems mounted on mobile platforms. As the vision systems are designed to function at longer operating ranges with relatively narrow field of views, the expectations from stabilization loops have increased in recent years. In order to design a good stabilization loop, high gain compensation is required. While satisfying high loop gains for disturbance attenuation, it is also required to satisfy sufficient loop stability. Structural resonances and model uncertainties put strict restrictions on achievable stabilization loop bandwidth for gimbaled vision systems. For that reason, satisfying high stabilization performance under modeling errors requires utilization of robust control methods. In this thesis, robust controller design in LQG/LTR, H-infinity and Mu frameworks is described for a two-axis gimbal. First, the modeling errors are found by investigating the locally linearized models under different conditions. Next, the performance indices and weights are determined by considering the allowable stabilization error and possible platform disturbance profile. Then generalized plants are obtained by using the nominal model and corresponding weights for three different design methods. Using these generalized plants, LQG/LTR, H-infinity and Mu controllers are synthesized. Stabilities and performances of the three designs are investigated in detail. After that, comparison of the controllers is made by investigating the robustness of corresponding closed loops. The thesis work is finished with the experimental studies and performances to validate the designed robust controllers.