Adaptive Observer-based Fast Fault Estimation of A Leader-follower
Linear Multi-Agent System with Actuator Faults
LIU Guosheng
1
, ZHANG Ke
1
, JIANG Bin
1
1. College of Automation Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China
E-mail: liuguoshengcys@126.com, kezhang@nuaa.edu.cn, binjiang@nuaa.edu.cn
Abstract: This paper concerns with the problem of adaptive observer-based fault estimation of linear multi-agent systems with
actuator faults. An undirected multigraph is used to represent the communication topology of a group of leader-follower vertical
takeoff and landing (VTOL) aircrafts, and one method of fast adaptive fault estimation (FAFE) using relative output estimation
error is proposed to achieve fault estimation for such multi-agent systems. Based on linear matrix inequality (LMI) technique, an
effective algorithm is designed to obtain the related parameters. Simulation results show the effectiveness of the proposed
method for multi-agent systems’ fault estimation.
Key Words: Multi-agent systems, fast fault estimation, adaptive observer, LMI
1 Introduction
Multi-agent systems have received increasing research
enthusiasm in both theory and application fields such as
mobile sensor networks, transportation systems, automated
highway systems, unmanned aerial vehicles, and multirobot
motion systems. During past several years, various scientific
communities in control and systems have paid compelling
attention to a series of fundamental problems of multi-agent
systems: consensus problems, swarm problems, cooperative
control, formation control, network optimization problems
and achieved a great deal of fruitful results in several
excellent books [1][2][3], a survey paper [4] and references
therein.
With the increasing scale and complexity of multi-agent
systems, how to autonomously operate those systems in a
safe and reliable way is undoubtedly becoming more and
more important and challenging. Hence, technology of fault
diagnosis, including fault detection (FD), fault isolation (FI)
and fault estimation (FE) of multi-agent systems is taken as
an attractive topic and receive considerable attention. For
details, see a survey [5].
Based on a geometric fault detection and isolation (FDI)
approach, [6] proposed a local/decentralized detection filter
for detecting faults in other spacecraft by determining the
required unobservable subspace of local system. Geometric
methodology for FDI was also applied in the paper [7].
Dynamic neural network-based FDI scheme for satellites
that are tasked to perform a formation flying mission was
studied in [8] and [9]. Besides, many scientific researchers
focus on observer-based FDI. Of particular relevance to this
note are the achievements [10-13]. Based on conventional
observer approach, the influence of network-induced delay
was evaluated in [10]. An observed-based residual generator
was well designed and used to model the residual dynamics
to be a discrete-time markovian jump linear system with
disturbance according to [11]. Reference [12] proposed a
novel integrate design approach of observer-based FD for
*
This work is partially supported by the National Natural Science
Foundation of China (61304112,61428303,61490703) and Natural Science
Foundation of Jiangsu Province (BK20131364).
LTI systems. In context of norm based residual evaluation,
the residual generator and evaluator were designed together
in an integrated form with the help of LMI technology. In
[13], process fault accommodation in a class of nonlinear
continuous-time systems was investigated, and a new fault
estimation module based on an adaptive estimator was
proposed. Then a fault tolerant controller was constructed to
compensate for the effect of the faults by stabilizing the
closed-loop system.
However, the capability of fault detection and isolation
without fault tolerance is not enough to achieve the safety
and reliability of multi-agent systems. In general, FDI is the
first step in fault accommodation to monitor the system and
determine the location of the fault, the final objective of
fault diagnosis is to obtain fault information, then design an
additive controller to compensate for the detected fault. In
such relationship, the technology of FE is considered as an
active and effective bridge as expected.
Nevertheless, the research achievement in this direction is
relatively less. In [14], an actuator fault isolation and
estimation scheme using a bank of repetitive learning
observers for a class of discrete-time nonlinear systems was
investigated, and sufficient conditions for the proposed FD,
FI and FE scheme were given. In the technical note [15], a
robust fault estimation approach, based on sliding mode
observers was proposed, considering a collection of agents,
exchanging relative information over a communication
network. More meaningfully, an observable subsystem was
created to design decoupled sliding mode observers.
Deeply motivated by present researching situation and
theoretical demands of fault estimation for multi-agent
systems, in this paper we focus on the problem of adaptive
observer-based fast fault estimation for linear multi-agent
systems with actuator faults. The main contributions of this
note lie in that one method of fast adaptive fault estimation
(FAFE) using relative output estimation error is proposed to
achieve the feasibility and rapidity of fault estimation for
multi-agent systems, on the other hand, based on LMI
technique, an novel and effective algorithm is designed to
calculate the related parameters .
The rest of the paper is organized as follows. Problem
statement including some preliminaries and background is
Proceedings of the 34th Chinese Control Conference
Jul
28-30, 2015, Han
zhou, China
6340