With the increasing interest in small, relatively inexpensive satellites, envisaged to be produced on a large scale, there is a need to minimise development costs by embodying as many common elements as possible. The research presented here is aimed at the production of a 'universal' attitude controller in partial fulfilment of this need. Continuous momentum exchange torque actuators are assumed at present. Such a controller must maintain a specified dynamic response to changing attitude demands through a wide range of spacecraft dynamics parameters and in the presence of external disturbance torques. The sliding mode approach is pursued in view of its relative simplicity and effectiveness, especially when dealing with the continuous non-linearities of a) the kinematic differential equations for large angle slewing manoeuvres and b) the spacecraft dynamics when the angular velocities are sufficiently high for gyroscopic inter-axis coupling to be significant. Since the pointing accuracy of any attitude control system is ultimately limited by the accuracy of the attitude measurement system, a minimal requirement is on-board instrumentation for high precision measurement of attitude angles. To avoid the necessity of additional instrumentation for the measurement of other state variables such as angular accelerations, the control system is observer based. The sliding mode controller embodies two novel features: a) control smoothing integrators to eliminate actuator drive chatter and b) block control principle to facilitate straightforward design despite the non-linearities. The control system may be designed simply by eigenvalue assignment to yield a specified closed-loop dynamic performance. Three possible special observers Transactions on the Built Environment vol 19, © 1996 WIT Press, www.witpress.com, ISSN 1743-3509 640 Structures in Space are investigated which are intended to be insensitive to spacecraft dynamics modelling errors, to be compatible with the control law The control system performance is demonstrated by simulation, including sensitivity to measurement noise.
Read full abstract