Management of continuous system models in DEVS-SCHEME: Time windows for event-based control

Abstract

This thesis describes the design and implementation of an extended knowledge-based modeling and simulation environment, in which the management of continuous-time models in DEVS-SCHEME is realized to meet the requirements of the modeling and simulation of a robot-managed laboratory aboard the forthcoming Space Station Freedom. The modular hierarchical modeling scheme is preserved in the continuous models by using DYMOLA, a continuous modeling language, as a bridge between the abstracted DEVS models and the continuous simulation language code (in DESIRE). Through operations on the System Entity Structure (SES), a knowledge representation scheme, models at different granularity levels are generated. Time-windows can be obtained by manipulating a pruned SES. These time windows can be used to generate an equivalent discrete-event model at a coarser granularity. Therefore, an event-based intelligent control strategy can be realized in this knowledge-based multi-facetted modeling environment. Continuous-time and discrete-event modeling and simulation can be merged with AI techniques.