How Things Work: Knowledge-Based Modeling of Physical Devices

Reference: Gruber, T. & Iwasaki, Y. How Things Work: Knowledge-Based Modeling of Physical Devices. KSL, August, 1990.

Abstract: The broad body of knowledge and tasks in engineering and physical science is of major applied interest to our governmental and industrial sponsors and of intrinsic interest to us. The textbook knowledge of engineering and physical science is usually well worked out and often has a strong theoretical basis. Other less formalized engineering knowledge, such as the assumptions and rationale that go into the design of an artifact, is extremely valuable. For these reasons, our research focuses on knowledge about engineered devices that describes their structure and how and why they work. In this research, we aim to contribute to the development of a comprehensive body of knowledge and reasoning methods in the realm of physics and engineering, embodied in a computationally useful form. We and our collaborators in universities and companies aim to provide a substrate of fundamental knowledge of physics and engineering upon which multiple systems can be built. Such a resource would support and motivate the development of techniques for modeling designed artifacts. General-purpose device models represent knowledge that can be used for several reasoning tasks, such as explaining how they work, why they are designed as they are, what might have caused a past failure, how they might break or wear out, and how they might be redesigned. Many of the tasks that comprise engineering practice could be enhanced by knowledge-based device-modeling technology. Making a variety of modeling methods accessible to engineers would facilitate the dissemination of knowledge to other engineers, and help capture the knowledge that goes into the engineering of devices. The engineering knowledge bases would serve as an institutional memory for engineering organizations, explicitly linking upstream tasks such as design and specification with downstream tasks such as manufacturing and maintenance.

Full paper available as hqx, ps.

Jump to... [KSL] [SMI] [Reports by Author] [Reports by KSL Number] [Reports by Year]
Send mail to: ksl-info@ksl.stanford.edu to send a message to the maintainer of the KSL Reports.