摘要：AbstractIn this paper, we present a novel dynamic consensus algorithm capable of tracking the maximum value of a given measurement in a distributed network. Each node in the network implements a sliding-mode based algorithm and only uses the information provided by its neighbors to track this maximum value on the network. Thus, at any given time, a network node is not allowed to disclose information from one of its neighbors to any other neighbor. We demonstrate the convergence and stability properties of this technique and provide a guide to select the control parameters, initial conditions, and sampling period for discrete-time implementations. From a practical perspective, the proposed technique is very promising since it relies on selecting only three parameters, which are the same for the whole network, and solving one numerical integration on each node. Numerical simulations illustrate the results and help visualize the algorithm transient behavior.
