Edinburgh Napier University

  The design of planar mechanical systems is a challenging and time-consuming task for engineers. Evolutionary computing has been shown to be a successful tool applied to design automation in various mechanical engineering fields. This thesis aims to determine how evolutionary computing can be applied in the early conceptual stage to support planar mechanical design. Building on existing work, it investigates suitable evolutionary representations and defines and evaluates a framework for evolving planar mechanical systems in a physics environment. It focuses on the design of representations in a multi-step approach, capable of producing mechanical shapes and mechanisms consisting of several components and linkages, adapting to their environment and able to traverse different landscapes.
Based on a review of the literature on evolutionary computing in design, shape representations, generative design tools, and evolving mechanisms, a generative system was developed, allowing a series of empirical studies to be conducted. Analysis of the results demonstrated the importance of breaking down the representation design into multiple stages. It showed that the implemented representations, combined with the generative tool and evolutionary operators, are capable of evolving solutions for problems with different complexity. The results indicate the representation’s large impact on the solution quality, and therefore, careful design is necessary. This work provides insight into design decisions and compatibility with evolutionary computing techniques, offering a promising outlook for using this method to support the conceptual stage of mechanical design. In future, this work has the potential to be developed into an industry tool for assisting engineers in the early stage of planar mechanism design.