Physics-Based Semantic Reasoning for Function Model Decomposition

Abstract

In graph-based function models, the function verb and flow noun types are usually controlled by vocabularies of standard classes. The grammar is also controlled at different levels of formalism and contribute to reasoning. However, the text written in plain English for the names of the functions and flows is not used for formal reasoning to help with modeling or exploring the design space. This paper presents a formalism for semantic and physics-based reasoning on function model graphs, esp. to automatically decompose black box models and to generate design alternatives using those plain-English texts. A previously established formal language, which ensures that function models are consistent with physics laws, is used as a baseline. Semantic reasoning is added to use the unstructured information of the flow phrases to infer possible means of decomposing the model into a topology connecting appropriate subfunctions and to generate multiple alternative decompositions. A data structure of flow nouns, flow attributes, qualitative value scales, and qualitative physics laws is used as the data representation. An eight-step algorithm manipulates this data for reasoning. The paper shows two validation case studies to demonstrate the workings of the language.