Intelligent switching between multiple model-based adaptive controllers using data-driven control theory

Adaptive control theory is a promising technology to improve the performance and stability of nominal controllers in the presence of system uncertainties. However, conventional model reference adaptive control schemes are not robust in the presence of unmodeled dynamics, actuator constraints etc. To overcome the above limitations, recently various new adaptive control schemes are proposed in the literature. Since the underlying mathematical constructs of these schemes are different, each adaptive scheme might be efficient in handling a particular class of unanticipated disturbance. Instead of depending on a single scheme to provide robustness, the best strategy will be to run all the adaptive schemes parallel and optimally switch between them depending on the current state of the system. This paper proposes a stable data-driven supervisory algorithm to switch between various model-based adaptive control schemes. The optimal switching is achieved using a data-driven control concept called `Unfalsified adaptive control'. The resulting data-driven collaborative adaptive controller greatly improves the performance and robustness of the nominal controller compared to independent adaptive schemes. Simulation results demonstrate the efficacy of the proposed algorithm.