Geometric data simplification for a virtual factory

Abstract

A virtual factory is a very useful model for analyzing complex manufacturing systems that cannot be described by analytical or mathematical models. One of the main obstacles in the simulation of a virtual factory comes from the huge size of the geometric data involved. This paper proposes an efficient method for reducing that data volume through introducing a simplification method for virtual factories that consist of prismatic solid models containing complex internal features. The proposed method allows a determination of whether the faces of a virtual factory represent internal features (invisible), which are unnecessary to the simulation and may therefore be eliminated. The proposed approach uses a graphics board optimized to render a given scene and reverses the process; after drawing a scene, we try to identify the entities that are visible. The proposed algorithm splits the given triangles into visible groups, and for each visible group, we can then draw the mesh on a pixel map from the corresponding viewing direction. By querying the colors that appear on the pixel map, it is possible to determine the visible triangles. When the visible triangles are known, it is possible to remove the internal features by removing those invisible triangles from the triangular mesh.