Model-based fault detection and diagnosis of selective catalytic reduction systems for diesel engines

Chen, Rui

Permalink

https://hdl.handle.net/2142/50387 Copy

Description

Title

Model-based fault detection and diagnosis of selective catalytic reduction systems for diesel engines

Author(s)

Chen, Rui

Issue Date

2014-09-16

Director of Research (if dissertation) or Advisor (if thesis)

Wang, Xinlei

Doctoral Committee Chair(s)

Wang, Xinlei

Committee Member(s)

• Hansen, Alan C.
• Lee, Chia-Fon
• Domínguez-García, Alejandro D.

Department of Study

Engineering Administration

Discipline

Agricultural & Biological Engr

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Diesel
• Selective Catalytic Reduction (SCR)
• Nitrogen Oxides (NOx)
• diagnosis
• On-Board Diagnostics (OBD)

Abstract

The continuously stringent emission regulations call for the adaptation of the Selective Catalytic Reduction (SCR) system by many diesel manufacturers. In order to show the EPA the latest emission compliance, the sensors are required to be installed at upstream and downstream of the SCR. As a result, the reduction efficiency is also required to be monitored by the On-Board Diagnostics (OBD) regulations. Specifically, a diagnostic algorithm is required to detect and isolate the SCR system faults that may cause emission violation. In this research, two model-based fault detection and isolation algorithms are developed to detect and isolate the dosing fault and the outlet NOx sensor fault for the SCR system. The dosing fault is treated as an actuator additive fault, while the outlet NOx sensor drift and/or offset fault is treated as a sensor additive fault. First, a 0-D SCR nonlinear dynamic model was developed to facilitate the model-based approaches. A parity equation residual generator was designed based on the linearized SCR model and the fault transfer function matrix. Then, a sliding mode observer based residual generator is designed directly based on the nonlinear model. The two diagnostic algorithms are then implemented in the Matlab/Simulink environment for validation. A high fidelity nonlinear 1-D SCR model is used to generate system outputs and to simulate the plant. The simulation results show that the two model-based fault diagnosis approaches are capable of detecting and isolating the outlet NOx sensor and dosing faults in real time.

Graduation Semester

2014-08

Permalink

http://hdl.handle.net/2142/50387

Copyright and License Information

Copyright 2014 Rui Chen

Owning Collections