Title :
Case-based planning for simulation
Author :
Alexander, P. ; Tsatsoulis, C. ; Holtzman, J. ; Minden, G.
Author_Institution :
Kansas Univ., Lawrence, KS, USA
Abstract :
The process of designing a complex system often involves the analysis or simulation of candidate designs via a computer simulation tool. The results of these simulations are heavily dependent on the expertise involved in the selection and use of simulation tools. SIMPLAN is an attempt to use a case-based approach to the generation of simulation plans. Each plan generated by SIMPLAN to solve a previous problem is stored as a case in the SIMPLAN case-base. In an attempt to access more implicit knowledge in a case, SIMPLAN represents a case using sequences of sub-cases. These sub-cases represent specific actions taken to realize the higher level case and are themselves complete cases. By indexing sub-cases as well as the higher level cases, SIMPLAN can build high level cases from the subcases. It is also felt that representing a plan in this manner makes it possible to interface the case-based planner with a traditional first principles planner, allowing a complete system to draw upon knowledge represented by experience in the case-base and knowledge of first principles represented in a classical knowledge-base
Keywords :
digital simulation; inference mechanisms; knowledge based systems; knowledge representation; problem solving; SIMPLAN case-base; automated planning; candidate designs; case-based approach; case-based planner; classical knowledge-base; computer simulation tool; expertise; higher level cases; implicit knowledge; simulation plans; sub-cases; traditional first principles planner;
Conference_Titel :
Expert Planning Systems, 1991., First International Conference on
Conference_Location :
Brighton