Robust Fault Diagnosis of the Mars Express Satellite

Dynamic Model Identification and Robust Residual Generation for the Fault Diagnosis of the Mars Express Satellite Attitude System. Seminar of two hours for the University of Science & Technology Beijing - USTB.
 

Abstract

• This seminar presents a model-based Fault Detection and Isolation (FDI) strategy for a satellite attitude system, where the process is only partially known and affected by significant disturbances and modelling uncertainty. Starting from input-output data, linear dynamic models are identified in discrete time using AutoRegressive eXogenous (ARX) and Errors-In-Variables (EIV) structures and then transformed into state-space innovation form. A key step is identifying the disturbance distribution matrix, so that the uncertain MEX system can be represented as a known model with unknown inputs. On this basis, diagnostic observers are designed to generate residual signals that are simultaneously sensitive to actuator and sensor faults and robust against noise, disturbances and model-reality mismatch. Fixed decision thresholds are derived from a geometric and statistical analysis of the residuals and then refined through Monte Carlo simulations to minimise false-alarm and missed-fault probabilities. The methodology is illustrated on the MEX satellite simulator, considering thruster and gyro faults under realistic disturbance conditions. The results show that the proposed integrated identification and residual-generation framework achieves reliable, robust fault detection and isolation, making it suitable for practical aerospace diagnosis applications.

Keywords

• Dynamic system identification; fault detection and isolation; residual generation; observer-based diagnosis; unknown-input systems; satellite attitude control; robustness; Monte Carlo analysis

Main Points

• Motivation for robust fault diagnosis in uncertain satellite attitude systems
• Data-driven identification of linear dynamic models from input-output experiments
• Design of observer-based residual generators robust to disturbances and modelling errors
• Statistical residual evaluation and decision thresholds to limit false alarms
• Application to the Mars Express satellite with actuator and sensor fault scenarios
• Discussion of limitations, practical implications and extensions to other spacecraft

Downloads: Talk Slide Files

• Seminar Slides (PDF file, 8 MB).

Selected References

• Patton, R. J., Uppal, F. J., Simani, S., & Polle, B. (2008). Reliable fault diagnosis scheme for a spacecraft attitude control system. Proceedings of the Institution of Mechanical Engineers, Part O: Journal of Risk and Reliability, 222(2), 139-152. DOI: 10.1243/1748006XJRR98
• Patton, R. J., Uppal, F. J., Simani, S., & Polle, B. (2010). Robust fault detection and isolation applied to thruster faults of a satellite system. Control Engineering Practice, 18(9), 1093-1109. DOI: 10.1016/j.conengprac.2009.04.011
• Patton, R. J., Uppal, F. J., Simani, S., & Polle, B. (2006). Monte-Carlo reliability and performance analysis of satellite FDI system. IFAC Proceedings Volumes, 39(16), 187-192.

Previous Page

Simani Home Page

Dipartimento di Ingegneria