Autonomy is an increasingly important technology for robotic scientific and commercial spacecraft. An important motivation for developing onboard autonomy is to enable quick response to dynamic environment and situations, including fault conditions that a spacecraft may encounter. The reliability of such autonomous capabilities depends on the quality of their knowledge of a spacecraft's health state. Model-based approaches to fault management, i.e. model-based fault diagnosis (MBFD), is one approach to continuously verify correct behavior in addition to diagnosing symptoms to estimate the spacecraft's health state. The proper functioning of MBFD is dependent on 1) the quality of the model that is analyzed and compared to the outputs of onboard sensors to estimate the system's health state, and 2) the correct functioning of the diagnosis engine that interrogates the model and compares its analysis to observed system behavior. We are currently developing Verification and Validation (V&V) approaches to provide the necessary confidence that MBFD systems are correctly estimating the health of on-board spacecraft components and systems. Our work is intended to narrow the gap between the rapidly maturing field of Model-Based System Engineering (including MBFD) and the less-well understood area of identifying and applying appropriate V&V techniques to MBFD. Our effort is investigating three areas: 1) developing V&V techniques for the diagnostic model, 2) developing V&V techniques for the diagnostic engine in isolation, and 3) developing V&V techniques for the diagnostic engine and model in combination. This paper describes the work we have completed in the first area. We describe our approach to selecting a system to be represented by a model, the approach to modeling the system, the verification approach we developed for the model, and the results of the verification activity. We conclude with a description of the work remaining in the last two areas, which will be addressed over the next two years.
[1]
Randall Davis,et al.
Diagnostic Reasoning Based on Structure and Behavior
,
1984,
Artif. Intell..
[2]
Ole J. Mengshoel,et al.
Verification and Validation of System Health Management Models using Parametric Testing
,
2011
.
[3]
Ksenia Kolcio,et al.
Model-based off-nominal state isolation and detection system for autonomous fault management
,
2016,
2016 IEEE Aerospace Conference.
[4]
Charles Pecheur,et al.
Survey of NASA V and V Processes/Methods
,
2002
.
[5]
Anuradha Kodali,et al.
Towards Accreditation of Diagnostic Models for Improved Performance
,
2004
.
[6]
E. A..
Simulation-Based Verification of Autonomous Controllers via Livingstone PathFinder
,
2004
.
[7]
Adam Sweet,et al.
Lessons Learned in the Livingstone 2 on Earth Observing One Flight Experiment
,
2005
.
[8]
Nicola Muscettola,et al.
Design of the Remote Agent experiment for spacecraft autonomy
,
1998,
1998 IEEE Aerospace Conference Proceedings (Cat. No.98TH8339).
[9]
Shin Nakajima,et al.
The SPIN Model Checker : Primer and Reference Manual
,
2004
.
[10]
Charles Pecheur,et al.
New V and V Tools for Diagnostic Modeling Environment (DME)
,
2002
.
[11]
L. M. Fesq,et al.
Integrating autonomous fault management with conventional flight software: a case study
,
1999,
1999 IEEE Aerospace Conference. Proceedings (Cat. No.99TH8403).
[12]
Gabor Karsai,et al.
A Tool Chain for the V and V of NASA Cryogenic Fuel Loading Health Management
,
2014
.
[13]
Lorraine M. Fesq.
Marple: an autonomous diagnostician for isolating system hardware failures
,
1993
.