Modular Representation of Components to Enable Generative Engineering

Generative design of engineering systems has been applied across various fields, including robotics, electronics, and architecture. Usually involving design grammars, a common challenge in all these approaches has been data and design representation. Design grammars also demand significant human input for development, are limited in scope, and require iterative testing to be fully applicable to a design problem. This paper introduces the concept of engineering component modularity, inspired by the modularity of words. The concept is presented in context of a generative design framework driven by a knowledge graph database. Additionally, a technical representation based on standardized practices is proposed, enabling the creation and simulation of software agnostic system architecture designs. This work is limited to the component representation, with the generation process itself, and the proposed capability to learn from the evaluated results, left for future research. The groundwork established here impacts the development of generative system design frameworks by enabling self-sustainable and robust autonomous design which requires minimal human inter-vention. It thus offers a tool for unbiased, efficient design space exploration. While previous research has addressed generative system design, the novelty of this approach lies in the use of a knowledge graph database in place of design grammar, combined with an integrated standardized simulation framework for more comprehensive evaluation.