Fault Detection for Linear Systems by Multiple Observers and Ellipsoidal Analysis

A new actuator fault detection approach is proposed in this paper for linear systems subject to unknown disturbance and noise. By considering that only one actuator in the system occurs fault at a time, limited number of fault modes are obtained. For each fault mode, an observer is synthesized to ge...

Full description

Saved in:
Bibliographic Details
Published in2024 IEEE 13th Data Driven Control and Learning Systems Conference (DDCLS) pp. 854 - 859
Main Authors Zhang, Wenhan, Gao, Lintao, Liao, Mingrui, Chen, Aijun, Li, Jintao, Wang, Zhenhua
Format Conference Proceeding
LanguageEnglish
Published IEEE 17.05.2024
Subjects
Actuators
Ellipsoidal analysis
Fault detection
Learning systems
Linear systems
Multiple observers
Noise
Observers
Residual evaluation
Sensitivity
Online AccessGet full text

Cover

More Information
Summary:A new actuator fault detection approach is proposed in this paper for linear systems subject to unknown disturbance and noise. By considering that only one actuator in the system occurs fault at a time, limited number of fault modes are obtained. For each fault mode, an observer is synthesized to generate the residual signal. Especially, the observer is optimized for certain mode to improve the system's sensitivity to faults. Based on the determined multiple observers, a residual evaluation and fault detection decision scheme are presented by the ellipsoidal analysis techniques. Numerical simulation of a VTOL aircraft system is utilized to demonstrate the validity of the presented scheme.
ISSN:2767-9861
DOI:10.1109/DDCLS61622.2024.10606898