An architecture of a knowledge-based simulation engine

The paper describes the architecture of a knowledge-based simulation engine (KBSE). It summarizes ongoing efforts at Knowledge Based Systems Laboratory, Texas A&M University and Knowledge Based Systems Inc., College Station, TX, in developing knowledge based simulation. The KBSE provides support for: developing a simulation model from a description of the system and the user concern in the form of a question to be answered; and analyzing the simulation output. The support for output analysis includes assistance for statistical analysis, explaining a set of observations, and generating alternative ways to improve performance/solve a problem. The approach used to develop KBSE was to integrate simulation expert knowledge with domain knowledge, commonsense reasoning techniques, and qualitative and quantitative simulation techniques. The Phase I KBSE automates the generation of simulation models from system descriptions. Phase II provides intelligent support for simulation output analysis. The paper presents a summary of the artificial intelligence (AI) based algorithms and heuristics developed in Phase I.