Experimental comparison between reaction wheel attitude controller strategies

Abstract

Systems for satellite attitude control are usually designed based on modern control techniques, which assume that the plant, sensors and actuators have linear behavior. A reaction wheel is an actuator that produces a torque as a function of electric current applied to a brushless DC motor. Ideally the output torque should be proportional to the current, but Coulomb and static frictions in the bearing introduce non linearities in the output torque, as function of the angular velocity of the wheel, especially at low speeds, near zero. These non linearities become more relevant specially in reaction wheels that are designed to rotate in both directions, causing the controller error to increase significantly at sense of speed reversals (zero crossings). This article presents results of a control orientation of an air bearing table by means of a reaction wheel. To check the controller action during a zero crossing, a small fan was attached to the table, producing a torque whose magnitude can be altered by adjusting the direction of the airflow. The control loop uses a fiber optic gyroscope (FOG) as angular velocity sensor. A PID digital controller drives the wheel based on the angular position of the table with respect to a given reference. In order to have a controller with a linear output torque, a mathematical model of torque was developed as function of the input current and wheel speed, from which it was constructed an algorithm of the inverse function, so that the non linearities were partially compensated (dynamic compensation). The results indicated that dynamic compensation can effectively reduce the maximum pointing error during zero crossing, while keeping constant other parameters such as response and settlement time.