Intelligent design assistance with SEA

This paper demonstrates how design assistance for engineering processes of computer based systems is realized with the SEA-Environment. SEA allows to specify engineering processes in an unambiguous way using extended Predicate/Transition Nets (Pr/T Nets) as underlying formal model. Since Pr/T Nets are executable, tool support for a design process is realized easily via simulation of the process specification. By linking a user defined abstract graphical representation to the Pr/T Net, the interface of a design assistant can be tailored to the needs and preferences of engineers. Furthermore, artificial neural networks are integrated into the SEA-Environment. Neural networks provide learning facilities, that allow learning how to take decisions during a design process, for instance the selection of an appropriate tool and its parameters among a set of alternatives, automatically. An advantage of this approach is that the huge amount of knowledge needed for various decisions does not have to be acquired a priori but can be gathered during the use of an engineering environment.