When executing parametric optimization using complex models of a large amount of CAD parameters, it becomes easier for geometry collapse to occur the more CAD parameters there are. This decreases the rate at which valid geometry can be generated. Additionally, a narrow range similarly narrows down the design space, wherein globally optimum solutions cannot be obtained.
JMAG uses GAs (Genetic Algorithms) to search for appropriate ranges of CAD parameters, and creates predictive models useful in estimating geometry validity prior to building the geometry through optimization. JMAG also possesses a function for creating the cases of a specified number of valid geometry based on the predictive model, and is capable of creating valid geometry in small amounts of time.
This document describes the method for using the functions useful in searching parameter ranges that reduce geometry collapse during optimization calculations.
Geometry collapse, Optimization, CAD Parameter, GA, Geometry, Predictive model, Dimension, Preinstall script, Python