LQG CONTROLLER DESIGNS FROM REDUCED ORDER MODELS FOR A LAUNCH VEHICLE

Abstract

The suppression of liquid fuel slosh motion is critical in a launch vehicle (LV). In particular, during certain stages of the launch, the dynamics of the fuel interacts adversely with the rigid body dynamics of the LV and the feedback controller must attentuate these effects. This paper describes the effort of a multivariable control approach applied to the Geosynchronous Satellite Launch Vehicle (GSLV) of the Indian Space Research Organization (ISRO) during a certain stage of its launch. The fuel slosh dynamics are modelled using a pendulum model analogy. We describe two design methodologies using the linear-quadratic Gaussian (LQG) technique. The novelty of the technique is that we apply the LQG design for models that are reduced in order through inspection alone. This is possible from a perspective that the LV could be viewed as many small systems attached to a main body and the interactions of some of these smaller systems could be neglected at the controller design stage provided sufficient robustness is ensured by the controller. The first LQG design is carried out without the actuator dynamics incorporated at the design stage and for the second design we neglect the slosh dynamics as well.