Towards an Evolvable Chromosome Model for Interactive Computer Design Support

In traditional function based evolutionary design synthesis, the encoding of the design is decided by the designer, and no formal process of encoding is provided in most research in this line. As a result of limitations in the encodings typically used, evolutionary design synthesis often produces only small-scale designs - that is, relatively few components are used. And because they fail to cover the full spectrum of user needs, and frequently do not consider manufacturing processes, costs and constraints, the design results are often not very useful to industry and only end up as 'new ideas' generated to satisfy academic curiosity. This paper tries to extend the algorithmic design process of traditional evolutionary synthesis approaches, integrating the product development process and broader customer needs. The method is to adopt a chromosome model based on the domain theory described in [1] to facilitate defining of encodings, e.g. to decide the most important parts and characteristics of the designs to be evolved. From the viewpoint of the chromosome model, this approach is an extension of the model towards an evolvable chromosome model within the domain theory. An evolvable chromosome model is reasonable because designs are intrinsically evolutionary and no design should be static. The evolvable chromosome model can also facilitate computer-aided conceptual design in an interactive evolutionary design system, thus pushing the research in functional-based evolutionary design synthesis a further step towards industrial-oriented applications.