Multiple-model based adaptive compensation of actuation sign uncertainty using an error transformation

This paper develops a new multiple-model based adaptive control scheme to compensate actuation sign uncertainty, which employs three steps: (1) design multiple estimators; (2) design multiple controllers using an error transformation relating to the output tracking error, the estimation error, and a virtual tracking error between the estimator state and the reference signal; and (3) design a control switching mechanism to select the most appropriate controller to generate the applied control signal. Such a control scheme is first deigned for single-degree of freedom bodies with a control gain that is unknown in both sign and magnitude, to introduce and illustrate this approach. Then, the full scheme is developed for three-axis rigid microsatellites with unknown inertia parameters and uncertain actuation signs. The proposed multiple-model based adaptive control scheme ensures the desired system stability and asymptotic tracking properties. Simulation results verify its effectiveness.