Session: 02-02-01: Computer-Aided and Simulation Driven Design 1

Paper Number: 166307

Achieving Integrated Top-Down and Bottom-Up Functional Synthesis by Clustering Solution-Neutral and Form-Independent Function Units Based on the Domain Synthesis Theory in Quotient Space 

Recent advances in generative artificial intelligence (GenAI) have been widely applauded as the precursor of a new paradigm of creative work, revolutionizing the creative field and demonstrating an unparalleled ability to generate innovative outcomes. However, only a limited number of GenAI tools support the initial stages of the creative design process due to the constraints and possibilities for these tools in exploring innovative ideas. One significant reason is design fixation, characterized as the unconscious adherence to a set of well-known ideas or knowledge that limits the scope of ideation, happening both consciously and unconsciously in all areas of creative work.

The initial conceptual designs, the early stages of new product development, play a crucial role in determining the cost and quality of products. In the functional decomposition and morphology (FDM) model, commonly used for developing conceptual design support tools, the conceived functional structures are closely linked to the behaviors of the matching physical structures or building blocks. As a result, tools based on the FDM approach inevitably encounter fixation issues, resulting in limited creative design capability and two significant drawbacks: a restricted ability to explore potential solutions and design fixation.

Maintaining a high level of abstraction for functionalities and employing both top-down and bottom-up functional search strategies are two ways to explore the conceptual design solution space fully. Employing form-independent and solution-neutral function units in function structures can significantly enhance the abstract representation of functionalities, leading to the generation of diverse alternative outcomes. In the combined top-down and bottom-up function decomposition processes, the top-down process breaks functions into sub-functions; in contrast, the bottom-up process reconstructs high-level functions from these lower-level sub-functions. As a result, a new search space is created through bottom-up functional synthesis, ultimately enhancing the creativity of the conceptual design outcomes produced during the functional reasoning phase.

To tackle the creative design challenges associated with design fixation and to enhance the capacity of computer tools for exploring more innovative outcomes, we present a function-form bidirectional mapping model. The sub-functions in the FBS framework have been further refined and expressed as fine-grained units termed segments of kinematic behaviors (SKBs). These units are solution-neutral and form-independent functional components, representing only segments of the mechanical system's kinematic behavior without necessitating embodiment in any specific mechanism or device. Thus, the sub-functions illustrated by the SKB subsets are unrelated to the behaviors of the implementing entities. As a result, design fixation in FDM-based functional synthesis approaches caused by the conceived functions having a close relation with physical solutions has been effectively addressed.

Applying the quotient space synthesis theory, the three-dimensional bottom-up SKB subsets clustering synthetic mode in the form layer is established. Through clustering of the SKB subsets coming from the multi-function behaviors of constituent mechanisms, a number of complete behavior groups are constructed to match the sub-functions in the function layer and to be realized by the multi-function behaviors of the constituent mechanisms in the form layer. In this way, a new way to innovate the constituent mechanisms through bottom-up clustering and reconstructing sub-functions depicted by the synthesized KSB subsets is proposed. While multiple functions of constituent mechanisms take part in the function-form mapping, a small number of constituent mechanisms can meet the complex functional requirements of design problems. In this way, the bottom-up SKB clustering approach proposed in this paper can generate novel functional principles and new functional schemes and guarantee that the sub-functions embodied by synthesized SKB subsets take place in time and space continuously and orderly. The creative design of mechanical systems based on synthesis SKB subsets will lay the groundwork for accelerating the application of Gen AI in the creative design of mechanical systems, promoting significant advancements similar to those in drug synthesis and related areas due to the introduction of GenAI.

Presenting Author: Yu-Xin Wang China University of Petroleum, East China

Presenting Author Biography: Yu-Xin Wang, male, born in 1964, doctor, graduated from Tianjin University (May 1994). He was a professor at Zhejiang University (2008-2012), Tongji University (2002-2007), and Tianjin University (1996-2001). He is now a professor at the China University of Petroleum Huadong. His primary interest is the creative design of mechanical systems and its automation, dynamic stability, and design of the axisymmetric vectoring exhaust nozzle (AVEN) for jets and energy-saving equipment. E-mail：strath_tj@hotmail.com

Authors:

Yu-Tong Li China University of Petroleum, East China
Yu-Xin Wang China University of Petroleum, East China