malin@nasa-jsc.CSNET (Jane Malin) (09/14/87)
A group here have been exploring combining qualitative modeling and discrete event simulation (rather than constraint propagation) methodologies for the purpose of analyzing system effects of component faults and failures. The purpose of such analysis would be to support failure modes And effects analysis and model-based development of fault management systems (including knowledge-based systems). The application area is space subsystems, especially of the process control type, such as power management and distribution, thermal control, or air and water purification. In this area, there is a need to model multiple operating modes of system components, including multiple failure modes. In fact, the system behaviors of most interest are changes from normal to faulty behavior that are consequences of faults and failures elsewhere in the system. We have a promising working prototype, which I reported on at the AI and Simulation workshop at AAAI-87. First, generally, I would like to know of others with ideas or experience in combining the two methodologies, and of technical problems encountered in their efforts. I would also like to get some ideas on approaches to the technical issues that arise from the need to model multiple modes, including failure modes, explicitly. Thanks in advance, Jane Malin. (malin%nasa-jsc.csnet@relay.cs.net).