A robust semi‐decentralized fault detection strategy for multi‐agent systems: An application to a network of micro‐air vehicles

Abstract

Purpose

The aim of this paper is to address the problem of fault detection (FD) of linear continuous‐time multi‐agent systems.

Design/methodology/approach

A mixed H_∞/H₋ formulation of the FD problem using semi‐decentralized filters is presented.

Findings

It is shown that through a decomposition approach the drawbacks of the existing distributed FD design methods in multi‐agent systems can be effectively tackled. An extended linear matrix inequality (LMI) characterization is used to reduce the conservativeness of the design solution by introducing additional matrices in order to eliminate the couplings of the Lyapunov matrices with the agent's matrices.

Research limitations/implications

It is shown that by applying the proposed decomposition approach the FD problem of multi‐agent systems can be solved by analyzing the problem of a set of decoupled systems whose order and complexity are equal to that of a single agent. This procedure will be useful for both simplifying the computational cost of the solution as well as for developing a fault detection filter having a semi‐decentralized architecture.

Practical implications

Application of this methodology to a network of micro‐air vehicles (MAVs) illustrates the effectiveness and capabilities of the proposed design methodology.

Social implications

The feasibility of the use of reliable and self‐healing network of unmanned systems, cooperative networks, and multi‐agent systems will be significantly enhanced and improved by the development of advanced fault detection and isolation (FDI) technologies.

Originality/value

A semi‐decentralized fault detection (FD) methodology is developed for linear multi‐agent networked systems to reduce the order and complexity of the observers at each agent. A mixed H_∞/H₋ formulation of the FD problem by using semi‐decentralized filters is presented. Using this approach each agent can not only detect its own faults but also is able to detect its nearest neighbor agents’ faults.

Citation

Davoodi, M.R., Khorasani, K., Talebi, H.A. and Momeni, H.R. (2013), "A robust semi‐decentralized fault detection strategy for multi‐agent systems: An application to a network of micro‐air vehicles", International Journal of Intelligent Unmanned Systems, Vol. 1 No. 1, pp. 21-35. https://doi.org/10.1108/20496421311298125