Matrix-based computational conceptual design with ant colony optimisation

Abstract

Early in the design process, few computational tools exist to help designers more fully explore a problem's solution space. Existing automated concept generators use knowledge from existing designs and the desired functionality of the new product to suggest solutions. These methods produce many candidate solutions, but often provide the designer with unmanageably large sets of solutions. So techniques are needed to organise the set of solutions into smaller groups. Under the premise that quality designs come from experienced designers, this paper presents a matrix-based automated concept generation method that is intended to analyse traditional conceptual design phase activities. The functional model is defined with the functional basis and a repository containing relationships of sub-functions and artificial components is developed. Morphological matrix, function–component matrix and design structure matrix are structured to produce and rank feasible concept solutions in the early design process. Furthermore, ant colony optimisation is used to optimise the generated conceptual candidate solutions and optimal results are demonstrated to the user automatically. Finally, a software prototype is developed and a case study showing the prototype applied to a design problem for transmission mechanism of loading platform is detailed.