0
Research Papers

Precedent-Free Fault Isolation in a Diesel Engine Exhaust Gas Recirculation System

[+] Author and Article Information
Michael E. Cholette

Department of Mechanical Engineering,  University of Texas at Austin, Austin, TX 78712cholettm@mail.utexas.edu

Dragan Djurdjanovic

Department of Mechanical Engineering,  University of Texas at Austin, Austin, TX 78712dragand@me.utexas.edu

The topological distance was computed using the Breadth-first procedure [27].

This is why σ(k) was reduced to narrow wm with each subsequent epoch.

www.tesis.de.

Registered Trademark of The MathWorks, Natick, MA.

J. Dyn. Sys., Meas., Control 134(3), 031007 (Mar 27, 2012) (11 pages) doi:10.1115/1.4005511 History: Received March 07, 2011; Revised October 13, 2011; Published March 27, 2012; Online March 27, 2012

In this paper, a recently introduced model-based method for precedent-free fault detection and isolation (FDI) is modified to deal with multiple input, multiple output (MIMO) systems and is applied to an automotive engine with exhaust gas recirculation (EGR) system. Using normal behavior data generated by a high fidelity engine simulation, the growing structure multiple model system (GSMMS) approach is used to construct dynamic models of normal behavior for the EGR system and its constituent subsystems. Using the GSMMS models as a foundation, anomalous behavior is detected whenever statistically significant departures of the most recent modeling residuals away from the modeling residuals displayed during normal behavior are observed. By reconnecting the anomaly detectors (ADs) to the constituent subsystems, EGR valve, cooler, and valve controller faults are isolated without the need for prior training using data corresponding to particular faulty system behaviors.

Copyright © 2012 by American Society of Mechanical Engineers
Your Session has timed out. Please sign back in to continue.

References

Figures

Grahic Jump Location
Figure 1

Voronoi Tessellation using seven weight vectors, ξi

Grahic Jump Location
Figure 2

Flowchart for the GSMMS training procedure utilized in this paper. The modification to the training procedure in Ref. [10] is emphasized by the dashed box.

Grahic Jump Location
Figure 3

Illustration of fault isolation through distributed anomaly detection. In this example, the fault is in subsystem 1.

Grahic Jump Location
Figure 4

Schematic of a generic EGR system

Grahic Jump Location
Figure 5

Block diagram of the EGR valve system

Grahic Jump Location
Figure 6

Characteristic curves for the EGR valve. These curves are plots of f2 in the valve model without (normal) and with four levels of obstruction (10%, 20%, 30%, 40%, and 50%) respectively.

Grahic Jump Location
Figure 7

The heat transfer coefficient for the EGR cooler for normal operation and four levels of fouling. As the fouling increases, the heat transfer coefficient decreases.

Grahic Jump Location
Figure 8

Block diagram of the EGR system with a delay fault inserted in between the controller and the valve

Grahic Jump Location
Figure 9

Simplified block diagram of the EGR system into a mass flow and temperature subsystem with the overall anomaly detection strategy

Grahic Jump Location
Figure 10

Diagram of the second level of anomaly detection. The overall AD (see Fig. 9) “splits” into a mass flow AD and temperature AD, each producing a corresponding CV.

Grahic Jump Location
Figure 11

The third level of anomaly detection utilized for isolation of EGR mass flow subsystem faults. Here, each AD consists of a GSMMS and a set of regional residual PDFs. Only the anomaly detector of the faulty subsystem(s) will show a low CV.

Grahic Jump Location
Figure 12

Driving profiles utilized for training and testing of the GSMMS models

Grahic Jump Location
Figure 13

Overall AD for various air flow obstructions in the EGR throttle valve for faults inserted at 1322 s

Grahic Jump Location
Figure 14

CVglobal for the second-level anomaly detectors. A valve anomaly introduced at 1322 s. Faults 1 and 2 are difficult to distinguish from the normal condition. This is due to the fact that the mass flow AD is monitoring the closed-loop system and the controller is able to compensate for small faults.

Grahic Jump Location
Figure 15

Fault isolation using the third level of distributed anomaly detectors for a valve anomaly introduced at 1322 s

Grahic Jump Location
Figure 16

Overall AD for various cooler fouling conditions, which manifest themselves as changes to the cooler heat transfer coefficient

Grahic Jump Location
Figure 17

CVglobal for the second level anomaly detectors: a cooler anomaly introduced at 1322 s. As expected, the cooler AD CV is low.

Grahic Jump Location
Figure 18

Overall AD for various controller delays in the EGR throttle angle PI controller

Grahic Jump Location
Figure 19

CVglobal for the second-level anomaly detectors for a controller anomaly (time delay) introduced at 1322 s. The CV for the mass flow subsystem indicates an anomaly is present.

Grahic Jump Location
Figure 20

Fault isolation using the third level of distributed anomaly detectors for a controller anomaly (time delay) introduced at 1322 s. The CV for AD4 drops while other CVs remain high, indicating a fault in the PI controller. It can also be seen that the amount of CV reduction is proportional to the delay.

Tables

Errata

Discussions

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In