Automated image interpretation for integrated topology and shape optimization

Abstract.

A fully automated configuration optimization system (ACOS) for two-dimensional structures is proposed. The first stage of this system employs homogenization or material distribution methods to create a grey level image of a structure with minimum compliance. The second stage involves image interpretation that converts the grey level image to a parameterized structural model ready to undergo shape optimization. Several basic geometric shape templates are used to match holes of varied size and shape inside a structure. Two unique shape characteristics are defined and matching formulae that assign an appropriate template to a hole are provided. In the final stage, shape parameters are treated as design variables in a shape optimization that seeks a fine-tuned configuration and ensures satisfaction of all design constraints. Since human-machine interaction is not needed, ACOS increases the efficiency of the integrated topology and shape optimization for configuration designs. Three configuration design examples are provided.