Session
Technical Session VIII: Advanced Subsystems and Components I
Abstract
In this paper two nonlinear control techniques are developed and compared for the satellite attitude control problem. The first technique is a robust recursive nonlinear method using Euler angle formulation. This method is related to integrator backstepping as well as feedback linearization techniques. However, in this paper a different formulation is presented which overcomes some of the previous difficulties in applying backstepping to this problem by treating the three axis satellite system as a fully coupled set of second order systems. The technique produces a robustly stable controller, which meets desired performance, accounts for system nonlinear behavior, and is easily implementable in a set of feedback equations that can be computed in real time. The second technique is a learning control that updates the control input iteratively in order to enhance the transient performance of systems that are repeatedly executed over a fixed finite duration. It updates the control input by learning laws without the computation of system parameters and inverse dynamics of systems. The advantage of utilizing learning control schemes for attitude control systems is the enhancement of transient performance from trial to trial by taking advantage of the periodicity of the repeated system operation. By learning unknown parameters or time functions, the learning control can compensate nonlinear dynamics so that the desired performance can be achieved. The performance of both techniques is demonstrated for a satellite attitude tracking maneuver which represents a satellite in a circular orbit maintaining one face toward the earth while tracking simultaneous sinusoidal pointing commands in the other two axes. It is shown that the recursive controller provides the desired tracking performance with reasonable control effort, and that the learning control technique can be used to compensate for periodic external disturbances.
Systematic Design of Attitude Control Systems for a Satellite in a Circular Orbit with Guaranteed Performance and Stability
In this paper two nonlinear control techniques are developed and compared for the satellite attitude control problem. The first technique is a robust recursive nonlinear method using Euler angle formulation. This method is related to integrator backstepping as well as feedback linearization techniques. However, in this paper a different formulation is presented which overcomes some of the previous difficulties in applying backstepping to this problem by treating the three axis satellite system as a fully coupled set of second order systems. The technique produces a robustly stable controller, which meets desired performance, accounts for system nonlinear behavior, and is easily implementable in a set of feedback equations that can be computed in real time. The second technique is a learning control that updates the control input iteratively in order to enhance the transient performance of systems that are repeatedly executed over a fixed finite duration. It updates the control input by learning laws without the computation of system parameters and inverse dynamics of systems. The advantage of utilizing learning control schemes for attitude control systems is the enhancement of transient performance from trial to trial by taking advantage of the periodicity of the repeated system operation. By learning unknown parameters or time functions, the learning control can compensate nonlinear dynamics so that the desired performance can be achieved. The performance of both techniques is demonstrated for a satellite attitude tracking maneuver which represents a satellite in a circular orbit maintaining one face toward the earth while tracking simultaneous sinusoidal pointing commands in the other two axes. It is shown that the recursive controller provides the desired tracking performance with reasonable control effort, and that the learning control technique can be used to compensate for periodic external disturbances.
