VIRTUAL PROTOTYPING USING KNOWLEDGE - BASED MODELING AND SIMULATION TECHNIQUES

Knowledge-based modeling and simulation bridges the gap between ''conventional'' artificial intelligence implementations (such as expert systems) and more traditional computer-aided design techniques. We are currently developing software, whose primary function is to capture a user-input design specification and produce a ''virtual'' rapid prototvpe in the form of executable rule-based code. This code can then be exercised either as an interactive part of a hardware-in-the-loop testbed simulator or as a component of an object-oriented ''behavioral'' simulation environment. While the Phalanx Testbed is the immediate beneficiary of this work, the techniques described have a wide range of application in the modeling of conceptual design and performance characteristics. This paper describes the system architecture and software tools that we are applying to generate virtual rapid prototypes for use in the Phalanx Testbed. Particular attention is paid to defining the intelligent knowledge-capture mechanisms and model generation methodologies that we are using to translate design knowledge and performance requirements into rule-based simulations. The object-oriented programming approach to the merging of ''new'' data with previouslv-captured and stored data is discussed, and the issues of verifying and validating prototypes generated using such partiallv ''reengineered'' models are examined. An application currently in use as an investigative prototype for testbed development, a simple position controller servomechanism used to control the azimuth angle of a target-tracking sensor, is used to illustrate the process.