Core Concepts
The paper designs an adaptive tracking control scheme to achieve asymptotically optimal tracking for non-periodic reference signals in stochastic regression systems with multi-threshold quantized observations.
Abstract
The paper considers an adaptive tracking control problem for stochastic regression systems with multi-threshold quantized observations. Unlike previous studies that focused on periodic reference signals, this paper deals with non-periodic reference signals, which poses additional challenges in ensuring the designed controller satisfies the required conditions for convergence of the parameter estimation.
The key contributions are:
The paper designs two backward-shifted polynomials with time-varying parameters and a special projection structure to overcome the difficulties in ensuring the persistent excitation and uniformly bounded conditions for the non-periodic reference signals.
The proposed adaptive tracking control scheme not only ensures the convergence of the parameter estimates to their true values, but also achieves asymptotically optimal tracking for the non-periodic reference signals.
The mean square convergence speed of the parameter estimation can reach the optimal rate of O(1/k) under suitable conditions.
The paper first introduces the stochastic regression system model with multi-threshold quantized observations and the non-periodic reference signal. It then presents the adaptive control algorithm, which consists of a weighted conversion of the quantized observations, an online recursive projection-based parameter estimation, and the design of the control input using the certainty equivalence principle.
The main theoretical results establish the convergence and convergence rate of the parameter estimation, as well as the asymptotic optimality of the proposed adaptive tracking control scheme. Finally, a simulation is provided to verify the effectiveness of the proposed approach.
Stats
Sentence
This paper considers an adaptive tracking control problem for stochastic regression systems with multi-threshold quantized observations.
Different from the existing studies for periodic reference signals, the reference signal in this paper is non-periodic.
The adaptive tracking control law can achieve asymptotically optimal tracking for the non-periodic reference signal.
The proposed estimation algorithm is proved to converge to the true values in almost sure and mean square sense, and the convergence speed can reach O(1/k) under suitable conditions.
Quotes
Quote
"The main difficulty is how to ensure that the designed controller satisfies the uniformly bounded and excitation conditions that guarantee the convergence of the estimation in the controller under non-periodic signal conditions."
"This paper designs two backward-shifted polynomials with time-varying parameters and a special projection structure, which break through periodic limitations and establish the convergence and tracking properties."