This paper suggests dense and switched modular primitives for a bond-graph-based GP design framework that automatically synthesizes designs for multi-domain, lumped parameter dynamic systems. A set of primitives is sought that will avoid redundant junctions and elements, based on preassembling useful functional blocks of bond graph elements and (optionally) using a switched choice mechanism for inclusion of some elements. Motivation for using these primitives is to improve performance through greater search efficiency and thereby to reduce computational effort. As a proof of concept for this approach, an eigenvalue assignment problem, which is to find bond graph models exhibiting minimal distance errors from target sets of eigenvalues, was tested and showed improved performance for various sets of eigenvalues.
Kisung Seo, Zhun Fan, Jianjun Hu, Erik D. Goodman,