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

Capturing knowledge without an extra burden on engineers is one of the most challenging issues in the engineering domain. This study proposes a framework of real-time knowledge capture and systematic formalization for the model-based design (MBD), the systems design approach where a simulation is a key component. The framework records the screen of the modeling tool as video data and spoken annotation in which an engineer verbally explains the rationale of the modeling conducted with Modelica-based software commonly used for MBD. The captured modeling tool log and spoken annotation are divided into chunks, each corresponding to a design operation, e.g., defining functions, defining system components, setting model elements, setting model parameters, and running simulations. The format of the graphical Issue-Based Information Systems (gIBIS), the argumentation model for representing design rationale represents the logical structure of the design process. We define design operation templates at the system and subsystem levels, which act as rules to convert the modeling tool log into gIBIS format systematically. We conduct a design case of a spinach harvesting machine with MBD software to verify the proposed framework. The results show that through non-intrusive recording of the design process, the proposed framework is effective in capturing knowledge that can be used to help understand how a model was built.