A multiple-model adaptive control scheme for multivariable systems with uncertain actuation signs

Abstract

A new multiple-model adaptive switching control scheme, using a direct adaptive control approach to avoid control gain estimate singularity, is developed for multivariable systems with uncertain actuation signs. Such an adaptive controller is capable of handling the adaptive control design condition uncertainty caused by actuation sign uncertainty. Multiple direct adaptive controllers are designed for all possible patterns of actuation signs, and each of them is updated by an adaptive law corresponding to one particular actuation sign pattern. A control switching mechanism is designed with multiple modified performance indexes based on normalized estimation errors, which is desirable for selecting the best controller. An aircraft flight control example is presented to show the effectiveness of the proposed adaptive control scheme.