Space Systems: A Software Engineering Approach towards Validation and Verification of Expert Systems

   Capabilities Being Demonstrated:
  • Verification Issues
  • Maintenance Issues

This project was accomplished as an on-site contract work at NASA Langley, Systems Validation Methodology Branch (SVMB)¹. A rule-based FDIR (Fault Detection, Isolation & Recovery) system, MMU (Manned Maneuvering Unit), developed by McDonnell Douglas Astronautics was used as a test bed. The MMU system was representative of the types of equipment commonly needed for space applications. Manned Maneuvering Unit is a backpack assembly to be used for untethered extra-vehicular activity in space. Since these systems cannot be tested in the weightless space environment, and there are no experts familiar with the common failure modes and effective diagnostic procedures, such knowledge-based systems have to be developed based on the analysis of the design of the system. The MMU expert system encoded a decision table enumerating possible failures of the MMU. This brute force approach of enumeration of cases into a flat set of rules made the rule base very difficult to comprehend.

Using the MVP-CA approach, the MMU system was analyzed based on development of a software model for the components of the system. Different structuring strategies were devised for this rule base which was able to group different thruster configuration rules, failure recovery rules and tank/thruster rules into separate groups semi-automatically. The modularization schemes devised made the knowledge base more comprehensible and maintainable for Space Station applications.

Another project in the area of Space Systems was an SBIR Phase I award from NASA Johnson. The focus of this project was to investigate the possibility of providing multiple structurings for ONAV (Onboard Navigation Expert System). ONAV was a CLIPS-based expert system built by NASA Johnson, and deployed on the space shuttle, to navigate the shuttle's re-entry. The 16 files provided a legitimate partitioning of ONAV according to the control aspects of the problem.

An initial analysis of ONAV indicated that partitioning a knowledge base according to control aspects of the problem was not sufficient for understanding the problem. The value of generating alternate partitioning schemes for ONAV was successfully demonstrated with a rudimentary version of the tool. Very subtle interrelationships were exposed across sets of rules in different files of ONAV that proved very valuable for maintenance, management and correct functioning of such a knowledge base.

  1. M. Mehrotra and C. Wild. Analyzing Knowledge-based Systems Using Multi-ViewPoint Clustering Analysis. Journal of Systems and Software 29:235-249. Special Issue on Software Quality in Knowledge-Based Systems, 1994.

  2. M. Mehrotra. Requirements and Capabilities of the Multi-View Point Clustering Analysis - In IJCAI-95 Workshop Notes on Knowledge-based Systems Verification, Validation and Testing, Montreal, Canada, August 1995.

  3. M. Mehrotra and C. Wild. Multi-View Point Clustering Analysis - In Proceedings, 1993 Goddard Conference on Space Applications of Artificial Intelligence, pp 217-231, May 1993.

  4. M. Mehrotra and C. Wild. Multi-View Point Clustering Analysis - Notes of Invited Workshop on Validation, Verification and Testing of Intelligent Systems in the 9th IEEE Conference on Artificial Intelligence for Applications, March 1-5, 1993, Orlando, FL.

  5. M. Mehrotra. Rule Groupings: An Approach towards Verification of Knowledge-Based Systems - Proceedings of Second CLIPS Users Group Conference, September 1991.

  6. M. Mehrotra. Rule Groupings: A Software Engineering Approach towards Verification of Expert Systems - NASA Contractor Report 4372 Contract NAS1-18585 May 1991.

  7. M. Mehrotra and S. C. Johnson. Rule groupings in expert systems - Proceedings of First CLIPS Users Group Conference, August 1990.

  8. M. Mehrotra and S. C. Johnson. Importance of Rule Groupings for Verification of Expert Systems - AAAI-90 Workshop Notes on Knowledge-based Systems Verification, Validation and Testing, Boston, August 1990 (selected for presentation also).

« Back | Top

¹ This work was done while Pragati's President, Ms. Mala Mehrotra was an employee of Vigyan, Inc.
2007 Pragati Synergetic Research, Inc. Research Focus | Products | Projects | Publications | Market Sectors | About Us