Session: CIE-15/16: SEIKM Joint Topics

Paper Number: 112226

112226 - A Framework of Real-Time Knowledge Capture and Formalization for Model-Based Design With Spoken Annotation and Design Operations 

Capturing design knowledge without an extra burden on engineers is one of the most challenging issues in the engineering domain. Previous studies have proposed successful knowledge capture methods by real-time logging in virtual engineering tools such as 3D CAD. However, few efforts have been taken to knowledge capture of the model-based design (MBD), the simulation-based system design approach. There are two open issues. The first is that knowledge of MBD includes modeling rationale considering focused phenomena, design conditions, the appropriate level of details, and so on. This specialized expertise usually remains in an engineer's tacit realm. It is not explicitly contained in the log of engineering tools. The second is that the MBD process is not straightforward but back and forth between system-subsystem levels and possible alternatives. Therefore, it must be more accessible for engineers to find helpful knowledge from the log if it is formalized according to the logical structure of the design process.
This study proposes a framework of real-time knowledge capture and systematic formalization for the MBD process. The framework records the screen of the modeling tool as video data. Character recognition generates the text log of the modeling tool, such as saving/loading files, selecting icons and inputting numbers, from the video. The two issues stated above are solved as follows. For the first issue, the framework records spoken annotation in which an engineer verbally explains the modeling rationale accompanying the tool operation log. For the second issue, the framework divides the tool operation log and spoken annotation into chunks so that every chunk corresponds to a design operation, such as defining functions, defining system components, setting model elements, setting model parameters, running simulations, and so forth. We represent the system structure with a function-structure mapping. Then, we adopt gIBIS, the argumentation model for representing design rationale, to represent the logical structure of the design process. We define design operations templates at the system and subsystem levels. They work as rules to convert the tool operation log into gIBIS format.
In the proposed framework, an engineer is just asked to conduct the MBD process with a simulation tool as usual and to explain his/her operations through the MBD process verbally. In this study, the authors semi-automatically conduct the knowledge formalization steps using the defined design operation templates after the log of the MBD process is captured.
As an example, we take a design case of a spinach harvesting machine conducted with Simulation X, the CAE software commonly used for MBD. A modeling expert took about 18 hours to conduct the design case. During the design process, the engineer built a model to simulate the harvester's performance in gripping, cutting, holding and storing spinach. Using the model, the engineer designed appropriate mechanisms. We applied the proposed framework to video and audio data of its modeling process to capture and formalize knowledge. Through the case study, we verified that the proposed method could capture and formalize engineering knowledge with the gIBIS format. We also interviewed the engineer about the burden of spoken annotation. Modeling while speaking increases the time required for modeling by 10-30%. Even so, since the knowledge could be explicitly recorded by speaking, the increased time was meaningful for the engineer. According to the results, the proposed framework is effective in the knowledge capture of MBD. This paper also discusses the open issues and our future work.

Presenting Author: Yutaka Nomaguchi Osaka University

Presenting Author Biography: Dr. Yutaka Nomaguchi is an associate professor of the department of mechanical engineering, Osaka University. He took Ph.D. at the department of precision machinery engineering, the University of Tokyo in 2002. His research topics include systems engineering, model-based design, design knowledge management and design creativity.

Authors:

Yutaka Nomaguchi Osaka University
Hiroyuki Morikawa Osaka University
Rempei Nishida Osaka University
Kikuo Fujita Osaka University