Fault detection in nonlinear stable systems over lossy networks

Document Type

Article

Publication Date

4-1-2013

Publication Title

IEEE Transactions on Control Systems Technology

Volume

21

Issue

6

First Page

2129

Keywords

Fault diagnosis, IEEE 802.15.4 wireless sensor networks (WSNs), Networked control systems (NCSs), Nonlinear systems, Quantized systems

Last Page

2142

Abstract

This paper addresses the problem of fault detection (FD) in nonlinear stable systems, which are monitored via communications networks. An FD based on the system data provided by the communications network is called networked fault detection (NFD) or over network FD in the literature. A residual signal is generated, which gives a satisfactory estimation of the fault. A sufficient condition is derived, which minimizes the estimation error in the presence of packet drops, quantization error, and unwanted exogenous inputs such as disturbance and noise. A linear matrix inequality is obtained for the design of the FD filter parameters. In order to produce appropriate fault alarms, two widely used residual signal evaluation methodologies, based on the signals' peak and average values, are presented and compared together. Finally, the effectiveness of the proposed NFD technique is extensively assessed by using an experimental testbed that was built for performance evaluation of such systems with the use of IEEE 802.15.4 wireless sensor networks (WSNs) technology. In particular, this paper describes the issue of floating point calculus when connecting the WSNs to the engineering design softwares, such as MATLAB, and a possible solution is presented. © 1993-2012 IEEE.

DOI

10.1109/TCST.2012.2236839

ISSN

10636536

Link to Full Text

Share

COinS