ABSTRACT In this articles, a novel distributed adaptive parameter estimation/identification algorithm for multi‐agent system (MAS) architectures is proposed, where parameter convergence is ensured under a relaxed mathematical condition called cooperative initial excitation (C‐IE). In this article, a MAS architecture is modeled as a strongly connected digraph network incorporating interagent communication delays. Classical distributed adaptive identifiers require a restrictive mathematical condition called cooperative persistence of excitation (C‐PE) for parameter convergence. The C‐PE condition is restrictive in contrast to the C‐IE since it requires the excitation/richness of information over the entire time span, unlike C‐IE needing the excitation only in the initial time span. The proposed algorithm employs a recently introduced weighted integrator dynamics, eliminating the need for computationally complex multiple switching mechanisms found in past literature while still ensuring parameter convergence. It guarantees uniform global exponential stability (without communication delays) and uniform global stability with asymptotic convergence (with communication delays) of the origin of the parameter estimation error dynamics, respectively. Simulation results further validate the efficacy of the proposed algorithm when compared with the state‐of‐the‐art methods. To the best of the authors' knowledge, this is the first work on relaxed excitation based distributed adaptive identification over a strongly connected digraph, offering stability guarantees in the presence of communication delays.
Garg et al. (Tue,) studied this question.