Search

As it is well-known, system uncertainties and unmodeled dynamics can deteriorate stability properties of model reference adaptive control systems. Motivated by this standpoint, we first analyse sta...

Summary Discrete‐time adaptive control algorithms can be executed directly in embedded code unlike their continuous‐time counterparts, which require discretization. However, their designs predicate...

A reference model of an adaptive control law defines how a closed-loop dynamical system under consideration has to asymptotically (or approximately) behave in the presence of system uncertainties. ...

For uncertain dynamical systems with actuator dynamics, this paper presents a new adaptive control architecture using expanded reference models. Specifically, the proposed approach allows the traje...

The objective of model reference adaptive control systems is to drive the trajectories of an uncertain dynamical system to the trajectories of a given reference model capturing a desired closed-loop system performance. To this end, most adaptive control signals take the form ua(t) = −ŴT(t)σ(x(t)), where x(t) ∈ ℝn denotes the state vector of an uncertain dynamical system, σ:ℝn → ℝs denotes a known basis function, and Ŵ(t) ∈ ℝs×m denotes an estimation of the unknown weight matrix W ∈ ℝs×m satisfying sm update laws (here m denotes the number of control inputs). In this paper, we focus on a class of reduced-order, computationally less expensive, model reference adaptive control systems that are only predicated on a scalar update law. Specifically, our contribution is to utilize a command governor architecture in order to improve transient performance of this class of adaptive control systems. We prove the stability of the overall closed-loop system using Lyapunov stability theory and we also present an illustrative numerical example for demonstrating the efficacy of the proposed architecture.

In the execution of networked control systems, a critical task is to reduce wireless network utilization while achieving closed-loop stability in the presence of system uncertainties. Motivated by this standpoint, we propose a new event-triggered model reference adaptive control architecture predicated on the hedging approach to schedule the actuator data transmission from the proposed adaptive control law to the uncertain dynamical system. The hedging approach alters the trajectories of an ideal reference model to allow for correct adaptation while reducing the actuator data transmission. Specifically, we show that the difference between the trajectories of the uncertain dynamical system and the hedged reference model vanishes asymptotically using Lyapunov stability theory. Unlike existing event-triggered adaptive control methods addressing the same problem, we show with the proposed architecture, in a non-conservative way, how the trajectories of the uncertain dynamical system eventually stays close to a given ideal reference model, and it can also allow for parameter convergence when the closed-loop system is persistently excited. An illustrative numerical example is also provided to demonstrate the efficacy of the proposed architecture over existing event-triggered adaptive control methods.

We study how to schedule actuator data transmission in discrete-time networked model reference adaptive control systems for reducing wireless network utilization. We propose an event-triggered hedging approach introduced to the reference model, which is constructed by taking the difference between the sampled control signal sent to an uncertain dynamical system and the actual discrete-time control signal computed by the adaptive control algorithm. The advantages of the proposed approach include i) asymptotic convergence of the difference between an uncertain dynamical system state and the reference model state predicated on a logarithmic Lyapunov function-based stability analysis, ii) possibility of convergence of the estimated parameters to the unknown parameters when the closed-loop dynamical system is persistently excited, and iii) direct execution of the proposed approach in an embedded code. Note that i) and ii) are not trivial to achieve without introducing the proposed hedging approach to the reference model and iii) results from our discrete-time framework since existing continuous-time approaches require discretization that may result in loss of stability properties. Finally, an illustrative numerical example is provided for demonstrating the efficacy of the proposed approach.