Distributed Interval Estimation Methods for Multiagent Systems

This article focuses on the study of distributed interval estimation for networked multiagent systems. For a multiagent system, it has many agents that could build an interconnection topology. For a distributed interval observer, it has many subobservers designed for the corresponding agents. Each subobserver contains two kinds of observer gains: one is determined based on the traditional observer design method and the other one is designed by using the information from the neighborhood. To construct distributed interval observers, two methods are proposed in this article. The first method combines the H-infinity technique with reachability analysis. This method reduces the constraint of design conditions and improves the accuracy of the estimation. However, it needs additional computation. The second one employs the monotone system theory and has no requirement on computation. Finally, two examples are used to compare the two methods.