Abstract
The potential for placing large precision reflectors in space is currently being investigated through the use of inflatable membrane structures. Their capacity for reducing launch mass and stowed volume is being exploited. However, on-orbit performance will require an understanding of the various influences on the deployment, inflation, and service of the membrane, and the associated effects on reflector surface precision.
Nonlinear controllers developed to improve performance of such systems are often dependent on state estimation and parameter identification procedures. The existence of these procedures, within the control strategy, increases the size of the algorithms, limiting the system performance in real-time. The research presented has as a main objective to create an intelligent controller, based on feedback error learning, which is capable of extracting performance information from precision large membrane deployables, and subsequently using this information to achieve maximum surface precision.
This paper presents a method to spatially discretize a doubly-curved membrane model into N = m × n spring-mass-damper cells. A recursive algorithm is developed and used in a simulator to predict the surface profile of the membrane. Each cell is connected to a feedback error learning controller in order to extract local state estimations. Simulation results are then compared to finite element predictions. The discrete cell model is shown to be simple enough for real-time control strategies, and potential methods for sensing/actuating to close the loop are discussed.
